Center for sensing technologies

Center for Innovation and Business Development

Center for Information Technologies

Project aim: FlexiGroBots is an Innovation Action aiming to build a platform for flexible heterogeneous multi-robot systems for intelligent automation of precision agriculture operations, providing multiple benefits to farmers around the world.
&nbsp;
About the project: Despite the rising farmer investment in agricultural robots, most deployable robotic systems are meant to automate only specific tasks. The wide variety of tasks that need to be fulfilled in a single precision agriculture operation makes it extremely unprofitable to address its automation with task-specific robots. These challenges result in a lack of flexibility of current heterogeneous multi-robot systems that poses low returns on investment and high risks for farmers. In order to become cost-effective, heterogeneous multi-robot systems needs to become more flexible by employing more versatile (e.g. multi-task) robots which collaborate to accomplish complex missions.
FlexiGroBots proposes a Platform for developing heterogeneous multi-robot systems and applications which allows for i) more versatility by using the same robots for different observation and intervention tasks, in different missions, throughout the crop life cycle, ii) more cooperation between heterogeneous (ground and aerial) robots to accomplish more complex missions; iii) more valuable data to generate accurate insights into the fields, crops and robotics operations by combining data from IoT sensors, satellites and data collected by the robots; iv) more autonomy for real-time adaptation of mission plans as well as robot behaviour at the crop level, given operational conditions and real-time insights; v) more precision to carry out specific tasks in a very localized way, gaining accuracy and lowering costs.

Research Associate in the field of technical/technological sciences-electronics, telecommunications, and information technologies

Oskar Marko

Marija Kopanja is a PhD student of Informatics at Faculty of Sciences, University of Novi Sad. At the same faculty, she has previously finished bachelor and master studies in Applied mathematics during which she gained knowledge in the field of finance, as well as in the field of the numerical optimization and machine learning, that represent main areas of her scientific research. She attended ECMI Mathematical Modelling Week in 2018. During the studies she has been financially supported by Ministry of Education, Science and Technological Development of the Republic of Serbia. She received her master’s degree with the thesis about analysis of machine learning methods for cost-sensitive classification problems and their application on imbalanced data within medicine. Besides machine learning, her main research interest is use of the explainable artificial intelligence and deep learning, for application in agriculture and medicine. She is actively involved in several projects, including DRAGON.

Marija Kopanja

Maša Buđen works as a Junior researcher at Center for Biosystems within the BioSense Institute. She graduated and got her Masters Degree at Faculty of Agriculture in Novi Sad. She holds a Bachelors Degree in Phytomedicine and Masters Degree in Genetics, Plant Breeding and Seed Science. Right now she is a PhD student at the University of Belgrade, Faculty of Agriculture studying Crop Science. Maša Buđen has experience in working on scientific experiments in the field of agriculture. Her research interests are related to the application of new technologies in plant production.

Maša Buđen

Slobodan Birgermajer is research associate in the Center for Sensing Technologies at the BioSense Institute. Mr. Birgermajer received his PhD in microware electronics form the Faculty of Technical Sciences, University of Novi Sad, in 2019. He is currently working on development and application of the sensors in agriculture, food industry, and environmental monitoring, and he is actively involved in many H2020 projects such as ANTARES, IPANEMA SHealty, etc. His main research interest includes microwave passive devices, photonic crystals, microfluidic sensors, and 3D and 4D printing.

Slobodan Birgermajer

Mina Mirović is a junior reseacher at BioSense Institute. She holds a BSc in Mechatronics from Faculty of Novi Sad and MSc in Mechatronics, Robotics and Embedded Systems from London South Bank University.

&nbsp;

Mina Mirović

Goran Kitić works at the position of the Head of the Center for Sensing Technologies within the BioSense Institute. He received a PhD degree from the Faculty of Technical Sciences, University of Novi Sad, Serbia in 2016 on the subject of microwave soil moisture sensors. Goran Kitić developed experience in the field of different fabrication technologies (LTCC, thin&amp;thick film technology, microfabrication, PCB, wire bonding, machining, and 3D printing) and characterization methods (electron microscopy, EDS spectroscopy, UV/Vis spectroscopy, optical profilometry, electrical measurements). His research interests are related to the development of sensors applied in agriculture, food sector and environmental protection. Some of the developed sensor prototypes involve a multispectral sensor for plant status assessment, soil moisture sensor, a device for early mastitis detection, nitrates in water detector. Besides stated, he is heavily involved in the development of a robotic system for soil sampling and analysis. His research work resulted in seven international journal publications, and numerous papers published on international conferences. During his research career, he was involved in 13 international and national projects.

Goran Kitić

Đorđe Dragojević currently works as junior researcher at the Center for Information Technologies, Biosense Institute, University of Novi Sad. He holds an BSc in Mechanical Engineering from the Faculty of Technical Sciences, University of Novi Sad with bachelor degree thesis „Determining the roll-over limit and assesment of the lateral stability of heavy goods vehicles“. During his studies he was involved in several projects at Biosystems Engineering Department, tasked with agricultural machinery and equipment design. In a professional domain, for over 8 years Đorđe contributed as a mechanical designer in civil engineering and automotive industries on a variety of projects,domestic and international. Driven by curiosity he participated in providing practical solutions for engineering problems ranging from geophysical equipment design for Novilog in Novi Sad to military vehicles and transportation equipment design for LOHR Industrie in Bačka Topola.

Đorđe Dragojević

Damir Krklješ works as a Senior Researcher at the BioSense institute. He received PhD at the University of Novi Sad, Faculty of Technical Sciences, Novi Sad, Serbia, in 2016. He worked as a research and teaching assistant at the Faculty of Technical Sciences, teaching electronics, sensors and actuators and mechatronics. During that time he participated in many national and international projects, where his main research topics involved robotics (sensorized, two degree of freedom gripper. Sensorized robotic foot for ZMP measurement), force sensors (optical and force sensing resistors) and capacitive angular speed and position sensors, based on printed flexible substrates. Starting in 2013 he moved to the industry, where he worked for a few companies, mostly in a design service area for leading worldwide industrial partners. He was engaged both, as an engineer, and as a head of the department. During that period he was involved in a development of industrial products involving high-end LED lightning digitally controlled power converters, sophisticated devices for water pumps monitoring, power converters, etc. Damir Krklješ published over 30 journal and conference papers.

Damir Krklješ

Csaba Petes received his MSc degree in 2017 from the Faculty of Technical Sciences, Department of microprocessor systems and algorithms, University of Novi Sad, Serbia. His Master thesis: Remote control of moving vehicles using DTMF codes and the video signal on large and small distances.
Csaba received his BSc degree at the Faculty of Technical Sciences, Department of Microcomputer Electronics, where he graduated in 2014. Graduation thesis: Development and implementation of program "Paint" on embedded system. After his college, he founded his own company (Sept 2013), the Consulting Agency Starprofit which represented the Güttler GmbH in Serbia and other foreign companies on Serbian market, mostly in agriculture business.
In parallel he worked as a sales manager and as a teacher of the High School "Mihajlo Pupin" and Technical School in Bečej from October 2014 until September 2017. Now he works at the Biosense Institute on Embedded software and robotics systems. He is on more projects (Antares, MIMA, AgroboFood, WaQuMoS) .
In those projects, Csaba Petes take part primarily in the development of analog and digital electronics for read-out system, design PCB layouts, realization PCB prototypes, supporting the overall design of the robot, integration and signal processing.
Contact address: petes.csaba@gmail.com or chaba@biosense.rs

Čaba Peteš

Dajana Vujaklija (F) works as a Project manager at the Business Development department. She has a strong background in developing and coordinating dissemination and communication strategies, as well as overall marketing strategies to accelerate brand or project visibility. Having worked with international clients, from start-up teams to medium size companies, Dajana has gained considerable expertise in developing strategies for generating effective and positive organizational exposure. She is a social media expert with 9 years’ experience in developing and managing social media strategies and plans.
Dajana has been a part of BIOSENSE Institute since 2018, and she successfully manages Dissemination and Communication activities on H2020 projects. Her main responsibilities include the creation of long term strategies and plans for public presentation of projects’ activities and results to targeted audiences, content creation, social media management, and coordination of overall communication activities.

Dajana Vujaklija

Željana Grbović is a young researcher and PhD student in electrical engineering at University of Novi Sad, Faculty of Technical Sciences. She is highly interested in the application of machine learning algorithms and deep learning in the field of agriculture. Her beginnings in machine learning started 2016 when her team ranked in the top 5 among 52 teams in the prestigious competition SP Cup organized by IEEE. She received her master’s degree with the thesis about ear detection in wheat thermal images using deep neural networks. She has expanded the field experiment on the wheat involving images from mobile phones to provide early prediction. She has got funding from Innovation Fund, for The Proof of Concept for her PhD research topic. She is a guest researcher at Wageningen University and Research (WUR) in the group of Computer Vision and Agro-Food Robotics. She was involved in a Humistatus project about the prediction of spreading fungal infection on tomatoes based on hyperspectral images funded by the company Green Yard and WUR. She was a member of the team who tested PerClass MIRA Software for hyperspectral image processing before it was launched as public. Her main interests are image processing and application of transfer deep learning in plant phenotyping, food quality and safety. She is an active member of the artificial intelligence community and was invited speaker on several meetups and conferences where she promoted obtained results on phenotyping. She was a leader of a team that won third prize in the national competition „Space4Women“. Also, she led a team that participated at Alibaba Get Global Challenge final competition. She is actively involved in a few Horizon 2020 projects including ANTARES, Cybele, Dragon (serves as Dissemination Manager) and AgROBOfood and a few industrial projects with partners.

Željana Grbović

Prof. Dr. Vladimir Crnojević is the director and founder of BioSense Institute – R&amp;D institute for information technologies in biosystems, the only European Center of Excellence outside of EU. BioSense has been established in 2015 based on his idea that dates back in 2005 and the vision that information technology and biotechnology will meet in the future. Under his leadership and from only a few enthusiastic individuals BioSense has grown into respectable institution with almost 200 employees and more than 70 PhD holders. BioSense is well known as a very multidisciplinary institute which applies state-of-the-art technology like AI, robotics, nanotechnology and new materials in biosystems – agriculture, environment and medicine. Apart from excellent science BioSense Institute is very much devoted to the industrial cooperation through the creation of innovative solutions and strongly supports innovation development through accelerator activities and spin-off establishment. BioSense has given strong support to the digital transformation of the public sector and society in Serbia tthroug the development of various systems used primarily in agricultural sector.

&nbsp;

He is a professor of computer science at University of Novi Sad, Serbia. He studied Electrical Engineering and obtained his Ph.D. in image processing from the same university in 2004. Since then he has been involved in a research and development in the field of machine learning – nowadays popularly known as artificial intelligence. He was one of the pioneers in this field in Serbia which resulted in a number of new PhDs and various industrial projects. He acted as a director of Power, Electronics and Communications Department at Faculty of technical sciences, University of Novi Sad and served a member of the National scientific board for Electronics, Communications and Information Technology. He is a member of the National council for artificial intelligence of Republic of Serbia and Council of Europe  Committee on Artificial Intelligence.

&nbsp;

Prof. Dr. Crnojević is a part of the 10-Member-Group, prestigious group of experts acting as consultants to the Secretary General of United Nations in the filed of science, technology and innovation for Sustainable Development Goals (UN SDG). He has published more than 100 papers in renowned journals and conferences and acted as a reviewer for some of the most eminent journals in the field. His current interests are related to the role of AI in the achievement of UN SDGs and the future of humanity.

Vladimir Crnojević

Sanja Brdar is a senior researcher and head of the Centre for Information Technologies at BioSense Institute. She received a PhD degree from the University of Novi Sad in 2016. Prior to starting her PhD studies, she spent two years working in the software industry with a major in databases design and development. In 2010 she was awarded a ten-month Basileus fellowship for joining the Bioinformatics Laboratory at University of Ljubljana which set the foundations for the PhD thesis on data/knowledge fusion in bioinformatics. Currently, she is actively involved in many projects including H2020 Antares, Dragon, Bestmap and Flexigrobots and teaches two master level courses at Faculty of Sciences. Her research interests include machine learning, explainable AI and bioinformatics. She works on ensemble methods, data fusion, clustering, and predictive modelling with applications in biology, agriculture, environmental sciences and health. She enjoys participating in worldwide data science challenges and with the team of researchers made significant results: placed third in Nokia Mobile Data Challenge (2012), finalist of Orange, France, Data for Development Challenge (2013), finalist/winner/third place of Syngenta Crop Challenge (2016, 2017, 2019) and finalist of UNDP Depopulation challenge (2020).

Sanja Brdar

Predrag Lugonja is a researcher at BioSense Institute. He graduated from the Faculty of Technical Sciences, University of Novi Sad, Department of Power, Electronic and Telecommunication Engineering. He is currently in the final year of his PhD studies at the Faculty of Technical Sciences. He has been working in the fields of GIS and remote sensing

Predrag Lugonja

Oskar Marko, Ph.D. is the Head of the Center for Information Technologies and the Assistant Director for Innovation and Cooperation with the Economy at the BioSense Institute. His research focuses on applying advanced machine learning and evolutionary algorithms in agriculture. He spent the 3rd year of his undergraduate studies at City University London, where he did his final BEng project in signal processing. He led BioSense’s team that developed novel Big Data algorithms for yield prediction, seed distribution, and optimization of sowing strategies, which secured the 1st prize for BioSense at the Syngenta Crop Challenge and CGIAR Inspire Challenge. He is actively involved in many Horizon 2020 projects including Antares, Cybele, Dragon, and Flexigrobots, and industrial projects with partners such as Krivaja, MK Agriculture, Delta, and other farming and insurance companies.

Miloš Pandžić is a Junior Research Assistant whose interest includes several research fields: Earth observation, remote sensing, image processing and data analytics. He is currently enrolled as a PhD student of Geodesy and Geoinformatics at the Faculty of Civil Engineering, University of Belgrade, where he also obtained his master and bachelor degree. In 2015 he received scholarship from the Government of Austria (OeAD) and spent 4 months at TU Vienna where he analysed geometric quality of Sentinel-2 satellite multispectral images. Outcome of this work was his master thesis. After graduation he worked at the Finnish Geospatial Research Institute (FGI) on several projects funded by Academy of Finland and European Space Agency (ESA). His main occupation was segmentation and change detection in satellite imagery related to agricultural activity, urbanization and deforestation. In 2018 he joined BioSense Institute where he works on data fusion of optical and radar images for yield prediction, crop classification, crop monitoring, etc. He is involved in several national and international projects including Antares, Cybele, Dragon, PARSEC, agROBOfood.

Miloš Pandžić

Dr. Marko Panić is a research associate in the field of technical sciences and information technologies at the BioSense Institute. His doctoral dissertation is related to the development of algorithms for accelerated image reconstruction from incomplete measurements using a sparse representation of signals and statistical Markov random models. His research direction includes statistical modeling of multidimensional data obtained from different sensors. He is currently working on the analysis and processing of images obtained from hyperspectral and multispectral cameras using advanced machine learning methods with application in assessing the quality of fruits and vegetables. As a member of the BioSense team, he participated in Syngenta's "Crop Challenge" competition, which resulted in winning first place. He is actively involved in many projects from the Horizon 2020 call, including Antares, Cybele, Dragon and Flexirobots, as well as in projects with companies such as Krivaja, MK Agri and Delta Agrar.

Marko Panić

Dejan Pavlović is a researcher at BioSense Institute and a PhD student of System Control and Signal Processing at School of Electrical Engineering, University of Belgrade. His research interests cover signal processing and machine learning application, particularly focused on the field of precision agriculture. Dejan received his bachelor’s degree in Biomedical Engineering from the Faculty of Technical Sciences, University of Novi Sad and master’s degree in Electrical and Computer Engineering from School of Electrical Engineering, University of Belgrade. He is currently working in the field of data analytics developing an algorithm for the classification of multiple-behaviour patterns in cattle using Convolutional Neural Networks. He is actively involved in many projects, including Horizon 2020 projects, Antares and Dragon.

Dejan Pavlović

Branislav Pejak is a PhD student of Electrical Engineering at the Faculty of Technical Sciences, the University of Novi Sad, where he also received his BSc (2017) and MSc (2018) in the field of Biomedical Engineering. During his bachelor and master studies, he held a scholarship from the Ministry of Education, Science and Technological Development, Republic of Serbia. Since March 2018, Branislav has been working as a Junior Research Assistant at the Center for Information Technologies, BioSense Institute. His main research is data analytics on satellite images. Currently he is working on crop classification and crop yield prediction using time series of multispectral images and GAN machine learning framework. Within the project he is also working on many practical usages of ICT in agriculture such as smart soil sampling and fertilization, irrigation zone mapping, grain yield and protein content prediction etc. He is actively involved in many Horizon 2020 projects including Antares, Cybele, Dragon and AgRoboFood.

Branislav Pejak

Bojana Ivošević is working on utilizing Unmanned Aerial Vehicles (UAV) for efficient monitoring of croplands, freshwaters, and terrains. She hails from a background in biology and ecology. She initiated her PhD in Kyungpook National University, South Korea, where she gained 2 years of experience with UAVs. She continued to pursue her field of interest at the Faculty of Sciences, University of Novi Sad, Serbia. Her current research focus is to implement UAV technology in precision agriculture to calculate vegetation indices important for agricultural crop yield production, disease detection and crop management. She is also interested in enforcing UAVs for animal detection and counting, monitoring algae bloom, land cover classification, and seasonal effects on the ecosystems.

Bojana Ivošević

FlexiGroBots

zastava-svg

This project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No 101017111. 


Horizon 2020

BioSense’s research in the domain of imaging technologies relies on different devices for acquisition of data. They vary in the parts of the spectrum they cover and the number of bands in which they operate. These devices include visible near infrared (VNIR) and near infrared (NIR) hyperspectral cameras for lab research, and multispectral and thermal cameras which can be also used for the in-field purposes.

&nbsp;

Laboratory analysis is more accurate as hyperspectral cameras perform line scanning in more than 200 spectral bands and to further increase the reliability of measurements, they need to be calibrated using the reference sheets. However, high accuracy requires a large storage and very efficient image processing algorithms, since a hyperspectral image can take up more than 2 GB of space. For reduction of dimensionality we are selecting the most discriminative spectral bands, bands with high signal-to-noise ratio or applying the standard techniques such as PCA and ICA. Thermal and multispectral images also require pre-processing, but it is more focused on noise reduction using filtration. In the next step, depending on the problem, we are relying on feature engineering, clustering techniques and deep learning to segment the image into the regions of interest. Afterwards, depending on the problem, we are using the techniques of post-processing for final representation of results, or we are feeding the results into the machine learning models for higher-level analysis.

&nbsp;

<img class="alignnone wp-image-1492 size-large" src="https://api.biosens.rs/wp-content/uploads/2021/06/50-1024x544.jpg" alt="" width="1024" height="544" />

&nbsp;

When it comes to applications, we used these advanced cameras to develop algorithms for detection of fungal infections in vegetables and for revealing bruises in fruit days before visible symptoms occur, but they have also proven extremely valuable in ecosystems monitoring and yield prediction. These algorithms allowed us to get more information about the plants, estimate shelf-life of agricultural products, predict the crop yield and provide objective recommendations in post-harvest decision-making.

Hyperspectral, Multispectral and Thermal Imaging

Agricultural robotics is a field that is becoming more and more mature, with new solutions constantly reaching the market. This area has seen a significant growth due to the shortage in workforce, high costs of labour and tediousness and repetitiveness of the tasks. BioSense uses robots as moving platforms for gathering the data about the crop production. These platforms are equipped with both commercially available sensors and sensors developed by the Group for Sensor Design.

&nbsp;

The latest project we have been working on is concerned with soil in-field analysis. The proposed system moves through the field and perform soil sampling and soil parameters analysis (macronutrients, micronutrients, pH, EC) as well as georeferencing of the results of the analysis. The complete process is autonomous since it is possible to predefine the route of robotic platform on the plot using GPS coordinates and a 3D Real-Time-LIDAR system for reconstructing the surrounding area and avoiding any possible obstacles. As a result of the analysis, multilayer maps are generated where each layer represents one of the parameters obtained by the proposed system. Based on these maps, the farmer will be able to make the right decisions, optimize the production and make full use of the precision agriculture in terms of efficiency, yield increase and sustainability of agricultural production.

&nbsp;

<img class="alignnone wp-image-1493 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/51.jpg" alt="" width="672" height="422" />

Robotics

At Biosense Institute we are using quadcopter DJI Inspire 1 to gain better insight into crop production and plan and manage agricultural operations more accurately. The technology can collect very high-resolution imagery below cloud level, with much more detail than the satellite imagery. The datasets include images captured with RGB and modified NDVI camera, but we are also using Flir thermal camera to make temperature measurements over large areas of heterogeneous terrain. Unmanned aerial vehicles are easy to use, as most mapping and data-collection missions are now conducted autonomously, by feeding the shape file of the desired area into the dedicated software.

&nbsp;

For image stitching we use Pix4D and AgisoftPhotoscan software to generate high-resolution orthomosaics, vegetation index maps and DSMs (Digital Surface Models). These orthomaps are important layers of information and they can be integrated into GIS systems along with other layers derived from additional sources. These datasets are then processed and used for training of advanced machine learning algorithms.

&nbsp;

BioSense uses UAS (Unmanned Aerial Systems) to detect nutrient deficiencies, pest infestations, water stress or excess and other conditions that affect crop development and improve the management of crops. They are especially useful for fields of a few dozens of hectares, such as orchards or vineyards, as scanning large fields of arable crops requires a lot of time. UAVs have also proven as effective tools for environmental monitoring, as they can cover medium-sized habitats and tell us more about the type of surroundings that insects and mammals grow in. The use of UAVs revolutionized spatial ecology and agriculture and allowed us to get high-resolution data and make optimal decisions with high accuracy.

&nbsp;

<img class="alignnone wp-image-1853 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/slika5.png" alt="" width="775" height="444" />

NDVI map of sugar beet crop

Drone Imaging

BioSense is the location of a Sentinel Data Hub and we are an active user of products from the ESA’s Copernicus programme. Before the launch of Sentinel satellites, our research was based on Landsat 7 and 8 and RapidView images, but today, it is mainly focused on data coming from Sentinel 1 and 2 missions. Images from Sentinel 2 are coming in 13 spectral bands, approximately every 5 days, depending on the cloud coverage, while Sentinel 1’s synthetic aperture radar allows for monitoring even in the case of clouds. This ensures continuous high-resolution monitoring of the environment and agricultural production in particular, throughout the growing season.

&nbsp;

&nbsp;

<img class="alignnone wp-image-1495 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/52.png" alt="" width="960" height="600" />

&nbsp;

Since 2012, we have been producing land-cover maps for the Serbian province of Vojvodina, which distinguish fields of different crops, urban areas, forests and water bodies. This allowed us to map animal habitats, observe their change and capture the trends in crop rotation. Using the principles of data fusion, we reached the sub-pixel resolution using data sources with heterogeneous pixel grids. The second line of research is concerned with yield prediction. Based on the ground truth coming from field records, crowdsourced data and yield maps, we are training advanced machine learning models to estimate the yield. These models are based on more than 30 satellite vegetation indices, that incorporate the domain knowledge about the crops, and their accuracy is increased as plants progress through phenological stages. Vegetation indices are also excellent proxies for the damage assessment, as some of them are excellent proxies for damage caused by hail, drought or fire. This puts insurance companies among our partners, along with the government, research institutions and, nevertheless, the farmers, who are using our results to optimise the decisions concerning the time and amount of irrigation and distribution of fertilisers and pesticides.

&nbsp;

Satellite imaging

Big data analytics is based on heterogeneous data sources coming at different velocity and at a different resolution. Besides satellite images, BioSense’s AI group uses probes for soil electrical conductivity, handheld devices and information coming from agricultural machinery. Electrical conductivity of the soil is an excellent indicator of soil texture, compaction and chemical content and these parameters are scanned at various depths, depending on the setup and the plants of interest. Elevation maps are not less important, as they determine the flow of nutrients and often correspond to the yield, while the handheld devices, such as the BioSense’s Plant-O-Meter, are allowing for continuous growth monitoring.

&nbsp;

&nbsp;

<img class="alignnone wp-image-1486" src="https://api.biosens.rs/wp-content/uploads/2021/06/46-212x300.jpg" alt="" width="220" height="311" /><img class="alignnone wp-image-2435" src="https://api.biosens.rs/wp-content/uploads/2021/06/prostorna-212x300.jpg" alt="" width="220" height="311" />

<img class="alignnone wp-image-1489" src="https://api.biosens.rs/wp-content/uploads/2021/06/48-212x300.jpg" alt="" width="220" height="311" /> <img class="alignnone wp-image-1490" src="https://api.biosens.rs/wp-content/uploads/2021/06/9-212x300.jpg" alt="" width="220" height="311" />

&nbsp;

Using the techniques of data fusion, heterogeneous data are aggregated into a single dataset and used for the delineation of management zones. Zone delineation is done by applying clusterization algorithms on data, either in normalised or weighted form, depending on the relative importance of input features. These management zones have different properties between one another, but soil can be considered homogeneous within each of them. This allows for issuing recommendations concerning agricultural production separately for each zone, depending on its specifics. Such localised decision-making is the essence of the precision agriculture and it is also a necessary first step in determining the optimal sensor locations, which maximise their effectiveness.

Geospatial data analytics

With the occurrence of non-parametric machine learning models and especially with the rising popularity of deep neural networks, most of classifiers and regressors used today are essentially black boxes. Their accuracy is often unmatchable by any conventional model, and although we know the principles and mathematics behind them, we cannot directly infer the complex relationships between input and output features. Having an accurate model has a huge practical value, but in order to have richer knowledge about the problem and push the boundaries of science, we need to delve deeper into the black box.

&nbsp;

<img class="alignnone wp-image-1474 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/39-1.jpg" alt="" width="1187" height="1028" />

&nbsp;

There are a few techniques we are relying on in knowledge discovery and all of them are based on indirect observations of the machine learning models. One of them is the random permutation test, which estimates the feature importance based on the decrease in model accuracy in case of random permutation of its values in the dataset. Morris’s and Sobol’s uncertainty and sensitivity analyses on the other hand are telling us more about the possible outputs of the system and the interaction between the input parameters. Model explanation analysis even goes a step further. For each feature used in predictive model, through sampling-based approximation it gives the estimate of feature’s positive or negative contribution to the output, which enables us to direct the domain researcher to where it’s the most effective. These and other methods allowed us to determine the influence of various soil and weather parameters on the yield and to characterise drought resistant maize hybrids or varieties of soya with stable yield.

&nbsp;

<img class="alignnone wp-image-1476 " src="https://api.biosens.rs/wp-content/uploads/2021/06/40-1024x768.jpg" alt="" width="351" height="263" /><img class="alignnone wp-image-1477 " src="https://api.biosens.rs/wp-content/uploads/2021/06/41-1024x768.jpg" alt="" width="351" height="263" />

Knowledge discovery

The goal of BioSense’s AI Group is to get a comprehensive and accurate overview of biosystems, make phenotyping and plant breeding more efficient, understand molecular biology better and offer optimal and automatic decision-making to the farmer. Our research starts with the problem and requires innovative application of state-of-the-art algorithms and development of the novel ones. Machine learning algorithms extensively used in BioSense applications include Random Forest, SVM, Decision Trees, Ridge Regression, Spectral Clustering and many others including special family of machine learning algorithms- Deep Learning, and where the complexity of the problem goes beyond the available algorithms, BioSense researchers develop the novel ones.

&nbsp;

Example of novel algorithm is include weighted histograms regression (WHR). WHR is based on a voting mechanism, where training samples are voting for the output value of a test sample. The votes are weighted according to the similarity of features and used to form a histogram that essentially represents the PDF of the output, while the expectation of the PDF is essentially the predicted value. This algorithm proved very useful for yield prediction, where it outperformed random forest, SVM and artificial neural networks.

&nbsp;

&nbsp;

<img class="alignnone wp-image-1478 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/42.png" alt="" width="503" height="504" />

&nbsp;

&nbsp;

Another novel algorithm is integrative clustering based on Nonnegative matrix factorization. Combining results of individual clusterings together into an ensemble leverages evidence accumulation in order to improve the results of clustering. Individual clustering results are merged into a joint cluster membership matrix that is further factorized into matrices of encodings and basis. Algorithm can integrate clusterings stemming from different data sets, different data preprocessing steps or different subsamples of features or objects.

&nbsp;

This algorithm proved very useful for functional genomics. Proposed approach based on nonnegative matrix factorization can fuse diverse data sources and infer gene clusters with high functional enrichment and improved gene coverage.

&nbsp;

&nbsp;

<img class="alignnone wp-image-1480 size-large" src="https://api.biosens.rs/wp-content/uploads/2021/06/43-1024x419.png" alt="" width="1024" height="419" />

Machine Learning

Machine learning and domain expertise are often viewed as two competing approaches for solving problems in any area, including agriculture. Our strategy is to use the best from both worlds. Machine learning algorithms show excellent performance in finding hidden dependencies within the data, but there are already well-established theories that explain the natural processes and they have been proven in practice. We are fully exploiting this fact by incorporating the domain knowledge in our models. This is usually done in the process of feature selection and feature engineering.

&nbsp;

One of the examples of feature engineering, based on the domain expertise, was used for the selection of stress-tolerant maize hybrids in Syngenta Crop Challenge in Analytics 2019. Drought is a phenomenon related to longer period of time without precipitation accompanied with high temperatures and low air humidity. However, there is no strict mathematical formulation of drought and many indicators have been used in practice.

&nbsp;

Based on daily values of maximum and minimum temperature, precipitation and vapor pressure in several stages throughout the growing season, we designed features indicating to drought and heat stress at the specific environment. They included maximum number of consecutive dry days, low precipitation sum in a certain growth stage and vapor pressure deficit. In this problem, the engineered features helped us to determine the drought-resistant hybrids that should be planted to maximise the yield and minimise the risk of low income for the farmer, but they can be used in other as well. Other examples include engineering of meteorological features for prediction of airborne pollen concentrations, engineering of elevation features to better predict the flow of nutrients in the field and engineering of features that describe the animal behaviour for prediction of disease and mammary inflammation.

<img class="alignnone wp-image-1471 size-large" src="https://api.biosens.rs/wp-content/uploads/2021/06/38-1024x768.jpg" alt="" width="1024" height="768" />

<img class="alignnone wp-image-1473 size-large" src="https://api.biosens.rs/wp-content/uploads/2021/06/39-1024x768.jpg" alt="" width="1024" height="768" />

Feature Engineering

Deep learning is a rather new method of machine learning, that is due to its specifics and excellent performance in image processing often considered as a separate field within artificial intelligence. It works best with big datasets which have multi-dimensional characteristics, and there are a number of such datasets processed at BioSense.

&nbsp;

The first group of datasets are hyperspectral, multispectral and thermal images acquired in the lab or during the outdoor field experiments. Here, the the goal was to segment the images and recognise characteristic patterns within them, such as bruises and scratches on apples, infected tomato sepals or ears in images of wheat. Different architectures were tested with various parameters, which included the number of convolutional and fully connected layers, type of activation functions and position of dropout layers. They all were optimised for a particular problem, concerning the size of input images and the number of samples in the database.

&nbsp;

<img class="alignnone wp-image-1469 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/337.jpg" alt="" width="640" height="480" />

&nbsp;

The second group consists of airborne particle measurements acquired with Rapid-E, the ﬁrst automated particle detector based on laser-induced ﬂuorescence technology, which records scattered light and laser-induced fluorescence patterns, representing the shape and chemical footprint of the particles. This is one of the examples of deep neural networks (DNNs) that relied heavily on convolutional layers, due to their ability to deal with datasets fused from different data sources.

&nbsp;

In both applications, preprocessing is perhaps the most vital task, as the noisy and non-standardised inputs are very difficult to analyse. Also, an important thing to note is that every problem requires a customised solution, whose complexity corresponds to the richness of the dataset and whose accuracy justifies its learning and execution times.

Deep Learning

The most of BioSense’s research in Big Data is currently concerned with two types of datasets that require huge storage capacity and extremely efficient algorithms that will process them in the specified time requirements.

&nbsp;

The first of them are satellite images. Sentinel 2 satellites are releasing new 13-band images of any given point on Earth approximately every 5 days. These images come in 100x100 km2 tiles, with a 10 m resolution. Within Cybele project, for example, we are analysing satellite images of the whole Europe to classify soya fields, meaning that we need to go through hundreds of tiles and apply the classification algorithm. In such situations, we are parallelising the code to better suit the multi-core server architecture, speeding up its parts using Cython, but also adjusting the implementations to lower the computational complexity of the algorithm.

&nbsp;

Another type of data are call detail records of mobile phone users and they satisfy all the V’s of big data analytics. These records consist of lists of established calls, exchanged text messages and registered internet traffic and are enormous in size. An important issue that needs to be addressed is concerned with privacy and it is a huge challenge to extract useful information from anonymised or aggregated datasets. Also, although the data is semi-structured, the traditional RDBMS technology for data analytics is not applicable to datasets of this volume, so we are developing processing workflows using Apache family of technologies, such as Spark, Hadoop and Hive. They allow us to get a deeper insight into human behaviour and mobility patterns and infer the interaction between rural and urban areas.

&nbsp;

&nbsp;

<img class="alignnone wp-image-1465 size-large" src="https://api.biosens.rs/wp-content/uploads/2021/06/35-1024x578.jpg" alt="" width="1024" height="578" />

Big Data Analytics

FLEXIGROBOTS-Flexible robots for intelligent automation of precision agriculture operations

Centar za inovacije i razvoj poslovanja

Centar za senzorske tehnologije

Centar za Informacione Tehnologije

Cilj projekta: FlexiGrobots je inovaciona akcija čiji je cilj razvoj platforme za fleksibilne multirobotske sisteme koji će omogućiti inteligentnu automatizaciju procesa precizne poljoprivrede, pružajući višestruke koristi poljoprivrednicima širom sveta.
&nbsp;
O projektu: Uprkos sve većem ulaganju farmera u poljoprivrednu robotiku, većina raspoloživih robotskih sistema je osmišljena tako da automatizuje samo određene procese. Širok spektar zadataka koje treba ispuniti u jednoj operaciji precizne poljoprivrede čini krajnje neisplativim automatizaciju koja se oslanja na robote usmerene na pojedinačne procese. Ovi izazovi rezultiraju nedostatkom fleksibilnosti trenutnih heterogenih multirobotskih sistema, što predstavlja nizak povrat ulaganja i visok rizik za poljoprivrednike. Da bi postali isplativi, heterogeni multirobotski sistemi treba da postanu fleksibilniji kroz upotrebu svestranijih robota koji su u mogućnosti da izvršavaju vise različitih zadataka, ali i da međusobno sarađuju u obavljanju složenih procesa.
&nbsp;
FlexiGroBots će kreirati platformu za razvoj heterogenih multi-robotskih sistema i aplikacija koje omogućavaju i) veću svestranost upotrebom istih robota za različite zadatke tokom životnog ciklusa useva, ii) veću saradnju između heterogenih (prizemnih i vazdušnih) robota kako bi se obezbedilo izvršavanje složenijih procesa; iii) preciznije podatke koji će omogućiti tačniji uvid u polja, useve i robotizovane procese kombinovanjem podataka IoT senzora, satelita i podataka koje su prikupili roboti; iv) veću autonomiju za prilagođavanje planova različitih operacija u realnom vremenu, kao i za ponašanje robota na nivou useva, u odnosu na trenutne uslove i potrebe; v) veću preciznost u izvršavanju određenih zadataka na veoma lokalizovan način, čime se postiže tačnost i smanjenje troškova.

Naučni saradnik u oblasti tehničko-tehnoloških nauka- elektronika, telekomunikacije i informacione tehnologije

Marija Kopanja je student doktorskih studija Informatike na Prirodno-matematičkom fakultetu, Univerziteta u Novom Sadu. Na istom fakultetu je prethodno završila osnovne i master studije Primenjene matematike tokom kojih je stekla znanje iz oblasti finansija, ali i numeričke optimizacije i mašinskog učenja koje predstavljaju glavne oblasti njenog interesovanja. Učestvovala je na ECMI Matematičkoj nedelji modeliranja 2018.godine. Tokom studija je bila stipendista Ministarstva obrazovanja, nauke i tehnološkog razvoja, Republike Srbije. Master studije je završila na temi analize metoda mašinskog učenja za rešavanje klasifikacionih problema osetljivih na cenu greške i njihove primene na nebalansirane podatke u oblasti medicine. Pored mašinkog učenja, njeno istraživačko interesovanje je usmereno na objašnjivu veštačku inteligenciju i duboko učenje, sa primenom u poljoprivredi i medicini. Trenutno je aktivni učesnik nekoliko projekata, uključujući DRAGON.

Maša Buđen je zaposlena na poziciji Mlađeg istraživača u Centru za Biosisteme na Institutu BioSense. Završila je osnovne i master studije na Poljoprivrednom fakultetu u Novom Sadu. Osnovne studije je završila iz oblasti Fitomedicine, dok je na master studijama završila program iz Genetike, oplemenjivanja biljaka i semenarstva. Trenuto pohađa doktorske studije na Poljoprivrednom fakultetu Univerziteta u Beogradu iz oblasti Ratarstva. Maša Buđen ima iskustva u naučno-eksperimentalnom radu iz oblasti poljoprivrede. Njena istraživanja su vezana za primenu nove inovativne tehnologije u poljoprivrednoj proizvodnji.

Goran Kitić je zaposlen na poziciji Rukovodioca centra za senzorske tehnologije na Institutu BioSens. Doktorirao je na Fakultetu tehničkih nauka, na Univerzitetu u Novom Sadu 2016. godine na temu mikrotalasnih senzora vlažnosti zemljišta. Goran Kitić ima iskustva u oblastima raznih tehnologija fabrikacije (LTCC, debeloslojna i tankoslojna tehnologija, mikrofabrikacija, PCB, wire bonding, mašinska obrada i 3D štampa) i metoda karakterizacije (elektronska mikroskopija, EDS spektoroskopija, UV/Vis spektroskopija, optička profilometrija, električna merenja). Njegova istraživanja su u vezi sa razvojem senzora koji svoju primenu imaju u poljoprivredi, prehrambenom sektoru i zaštiti životne sredine. Neki od razvijenih prototipa senzora uključuju multispektralni senzor za utvrđivanje statusa biljaka, senzor vlažnosti zemljišta, uređaj za rano otkrivanje mastitisa, detektor nitrata u vodi. Pored navedenog, u velikoj meri je uključen u razvoj robotskog sistema za uzorkovanje i analizu zemljišta. Njegovo istraživanje je publikovano u 7 međunarodnih časopisa, i brojnim radovima objavljenim na međunarodnim konferencijama. Tokom svoje karijere, učestvovao je u 13 međunarodnih i nacionalnih projekata.

Slobodan Birgermajer je naučni saradnik u Centru za Senzorske Tehnologije pri Institutu BioSens. Doktorat iz mikrotalasne elektronike odbranio je 2019. godine na Fakultetu Tehničkoih Nauka, Univerziteta u Novom Sadu. Trenutno radi na razvoju i primeni senzora u poljprivredi, prehrambenoj industriji i monitoringu životne sredine, i aktivno je uključena na više međunarodnih projekata iz poziva H2020, kao što su ANTARES, SHEALTY, IPANEMA, i drugi. Takođe, domen istraživanja obuhvata razvoj i fabrikaciju mikrotalasnih pasivnih kola, fotonskih kristala, mikrofludinih senzora, kao i 3D i 4D tehnologiju štampe.

&nbsp;

Mina Mirović je zaposlena kao mlađi istraživač na BioSense Institutu. Osnovne studije iz oblasti mehatronike završila na Univerzitetu u Novom Sadu a Master studije iz oblasti mehatronike, robotike i embeded sistema završila je na London South Bank Univerzitetu.

Đorđe Dragojević je trenutno zaposlen kao mlađi istraživač u Centru za Informacione Tehnologije na Biosens institutu , Univerzitet Novi Sad. Diplomirao je na odseku Mašinstvo na Fakultetu Tehničkih Nauka, na katedri za Mehanizaciju i Konstrukciono mašinstvo sa usmerenjem Motori i Vozila sa temom završnog rada „Određivanje granice prevrtanja i provera bočne stabilnosti teških teretnih vozila“. Tokom studija učestvovao je na projektima u okviru Katedre za Biosisteme sa konkretnim učešćem u dizajnu poljoprivredne mehanizacije i opreme. U profesionalnom domenu, tokom više od 8 godina Đorđe je učestvovao na domaćim i međunarodnim projektima u građevinskoj i auto-industriji kao mašinski inženjer-konstruktor. Vođen radoznalošću doprineo je iznalaženju praktičnih rešenja inženjerskim problemima, dijapazona od dizajna merne opreme za geofizička istraživanja u okviru kompanije Novilog Novi Sad do dizajna opreme vojnih vozila i transportne opreme za potrebe vojske u okviru kompanije LOHR Industrie iz Bačke Topole.

Damir Krklješ radi kao viši istraživač na institutu BioSense. Doktorirao je na Univerzitetu u Novom Sadu, Fakultet tehničkih nauka, Novi Sad, Srbija, 2016. Radio je kao istraživač i asistent u nastavi na Fakultetu tehničkih nauka, predajući elektroniku, senzore i aktuatore i mehatroniku. Za to vreme učestvovao je u mnogim nacionalnim i međunarodnim projektima, gde su mu glavne teme istraživanja bile robotika (senzorisana hvataljka sa dva stepena slobode. Senzorisano robotsko stopalo za ZMP-metenja), senzori sile (optički i otpornici osetljivi na silu) i kapacitivni senzori ugla i ugaone brzine, na bazi štampane elektronike na fleksibilnim podlogama. Počevši od 2013. prelazi u industriju, gde je radio za nekoliko kompanija, uglavnom u oblasti dizajnerskih usluga za vodeće svetske industrijske partnere. Bio je angažovan i kao inženjer i kao šef odeljenja. U tom periodu bio je uključen u razvoj industrijskih proizvoda koji uključuju vrhunske LED rasvete, digitalno kontrolisane pretvarače snage, sofisticirane uređaje za nadzor pumpi, pretvarače snage itd. Od 2019 radi na BioSens institutu. Damir Krklješ objavio je preko 30 časopisa i konferencijskih radova.

Csaba Petes je 2017. godine završio Master akademske studije na Univerzitetu u Novom Sadu, na Fakultetu tehničkih nauka, smer mikroprocesorski sistemi i algoritmi. Tema njegovog maser rada je bilo daljinsko upravljanje pokretnih vozila pomoću DTMF kodova i video signala na velike i male daljine. Osnovne akademske studije završio je 2014. godine na Univerzitetu u Novom Sadu, na Fakultetu tehničkih nauka, smer mikroračunarska elektronika, sa temom diplomskog rada : Razvoj i primena programa „Paint“ na embeded sistemu.
U među vremenu (septembar 2013) osnovao je sopstvenu kompaniju, konsultantsku agenciju „Starprofit“ koja je zastupala „Guttler GmbH“ u Srbiji i druge strane kompanije na srpskom tržištu, uglavnom u poljoprivrednom poslovanju. Paralelno radi kao menadžer prodaje i kao profesor stručnih predmeta u Elektrotehničkoj skoli „Mihajlo Pupin“ u Novom Sadu i u Tehničkoj školi u Bečeju od oktobra 2014. do septembra 2017. godine.
Od maja 2018. godine radi na Institutu Biosense, na embeded softveru i robotskim sistemima. Radi na više projekata (Antares, MIMA, AgroboFood, WaQuMoS), prvenstveno na razvoju analogne i digitalne elektronike za sistem očitavanja, na dizajniranju izgleda PCB-a, realizaciji prototipa PCB-a, podržavajući celokupan dizajn robota, integraciju i obradu signala. Kontakt adresa: petes.csaba@gmail.com ili chaba@biosense.rs

Dajana Vujaklija radi kao menadžerka projekata Centra za inovacije i razvoj poslovanja instituta BioSens. Dajana poseduje dugogodišnje iskustvo u razvoju i upravljanju strategijama komunikacija, kao i dugoročnim marketinškim strategijama u cilju povećanja vidljivosti brenda ili projekta. Radeći za internacionalne klijente, od startap timova do srednjih preduzeća, Dajana je stekla značajnu ekspertizu u razvoju strategija koje doprinose građenju pozitivnog imidža organizacije. Dajana je ekspert za rukovođenje društvenim mrežama sa devetogodišnjim iskustvom u razvoju i upravljanju strategijama i planovima nastupa na različitim društvenim mrežama.
Dajana je deo tima BioSens instituta od 2018. godine od kada uspešno upravlja aktivnostima komunikacije i diseminacije na različitim H2020 projektima. Njena zaduženja uključuju kreiranje dugoročnih strategija i planova za javno predstavljanje aktivnosti i rezultata projekata ciljnim grupama, kreiranje sadržaja, upravljanje društvenim mrežama i sveukupna koordinacija komunikacionih aktivnosti projekata.

Željana Grbović je istraživač i student doktorskih studija na departmanu Energetika, elektronika i telekomunikacije na Fakultetu tehničkih nauka, Univerzitet u Novom Sadu. Njena oblast interesovanja obuhvata primenu mašinskog učenja u poljoprivredi. Algoritmima veštačke inteligencije je počela da se bavi 2016. godine kada je njen tim rangiran u TOP5 u konkurenciji od 52 tima na prestižnom takmičenju u obradi signala, SP Cup koji organizuje IEEE. Završila je master studije 2017. godine na temi detektovanja klasova pšenice u termalnim slikama korišćenjem dubokih konvolucionih neuronskih mreža. Istraživanje o pšenici je proširila i na ranu predikciju prinosa, na osnovu slika sa mobilnog telefona i drona, za šta je dobila finansiranje od Inovacionog Fonda, za Dokaz Koncepta. Gostujući je istraživač na Univerzitetu u Vageningenu (Wageningen University and Research (WUR)), u grupi za Kompijutersku viziju i robotiku u poljoprivredi. Učestvovala je u industrijskom projektu WUR-a i kompanije GreenYard, pod nazivom Humistatus, za predikciju širenja patogenih oboljenja na paradajizu, na osnovu hiperspektralnih slika. Glavni domen istraživanja je transfer dubokog učenja u oblasti procene prinosa, kvaliteta i bezbednosti useva u poljoprivredi. Predvodila je tim koji je osvojio nagradu na studentskom univerzitetskom takmičenju „Space4Women“ i koji je učestvovao u finalu svetskog „Alibaba Get Global Challenge“ takmičenja. Aktivni učesnik je na nekoliko Horizon 2020 projekata, uključujući ANTARES, Cybele, Dragon (Menadžer diseminacije) i AgROBOfood, kao i projekte sa kompanijama.

Prof. dr Vladimir Crnojević je direktor i osnivač Instituta BioSens – istraživačko-razvojnog instituta za informacione tehnologije biosistema, jedinog Evropskog Centra Izvrsnosti van EU. BioSens je osnovan 2015. na njegovoj ideji iz 2005. godine i viziji o spajanju informacionih tehnologija i biotehnologije u budućnosti. Pod njegovim rukovođenjem i sa nekoliko entuzijastičnih ljudi Institut BioSens je prerastao u respektabilnu instituciju sa gotovo 200 zaposlenih i više od 70 doktora nauka. BioSens je prepoznat kao veoma multidisciplinarna  institucija koja primenjuje najsavremenije tehnologije kao što je veštačka inteligencija, robotika, nanotehnologije i novi materijali budućnosti u biosistemima – poljoprivredi, životnoj sredini i medicini. Pored izvrsne nauke institut BioSens je veoma posvećen saradnji sa industrijom kroz stvaranje inovativnih rešenja i snažno podržava razvoj inovacija kroz akceleratorske programe i formiranje spin-of kompanija. BioSens je dao značajan doprinos digitalnoj transformaciji javnog sektora i društva u Srbiji kroz razvoj različitih sistema, primarno u sektoru poljoprivrede.

&nbsp;

Profesor je računarskih nauka Prirodno-matematičkom fakultetu Univerziteta u Novom Sadu. Na istom univerzitetu je studirao elektrotehniku i odbranio doktorat 2004. godine. Od tada je posvećen istraživanjima i razvoju u oblasti mašinskog učenja – danas popularne veštačke inteligencije. Jedan je od pionira u ovoj oblasti u Srbiji što je rezultiralo većim brojem odbranjenih doktorata i razvojem brojnih inovativnih rešenja za industriju. Obavljao je funkciju direktora departmana za energetiku, elektroniku i telekomunikacije Fakulteta tehničkih nauka Univerziteta u Novom Sadu i člana matičnog odbora za elektrotehniku, telekomunikacije i informacione tehnologije Ministarstva prosvete i nauke Republike Srbije. Član je nacionalnog Saveta za veštačku inteligenciju i Komiteta za veštačku inteligenciju Saveta Evrope.

&nbsp;

Prof. dr Crnojević je član 10-Member-Group, prestižne grupe eksperata koji imaju ulogu konsultanata Generalnog sekretara Ujedinjenih Nacija u oblasti uloge nauke, tehnologije i inovacija u realizaciji ciljeva održivog razvoja UN SDG koje su usvojile Ujedinjene Nacije. Objavio je više od 100 radova u renomiranim časopisima i na konferencijama i bio recenzent za neke od najeminentnijih časopisa u njegovoj oblasti. Njegova sadašnja interesovanja vezana su za ulogu veštačke inteligencije u ostvarivanju ciljeva održivog razvoja UN SDG i budućnosti čovečanstva.

Dr Sanja Brdar je naučni saradnik i rukovodilac Centra za informacione tehnologije na Institutu BioSens. Doktorirala je na Univerzitetu u Novom Sadu 2016. Pre početka doktorata provela je dve godine u softverskoj kompaniji u timu za razvoj baza podataka. Desetomesečnu Basileus stipendiju za istražvački boravak u Laboratoriji za bioinformatiku na Univerzitetu u Ljubljani dobila je 2010. godine i time postavila osnove za doktorat u oblasti fuzije podataka/znanja u bioinformatici. Aktivno je uključena u mnoge projekte, uključujući H2020 Antares, Dragon, Bestmap i Flekigrobots i predavač je na dva master predmeta na Prirodno-matematičkom fakultetu, smer Nauka o podacima. Njena istraživačka interesovanja uključuju mašinsko učenje, objašnjivu veštačku inteligenciju i bioinformatiku. Radi na ansambalskim metodama, fuziji podataka, klasterizaciji i prediktivnom modelovanju sa primenom u biologiji, poljoprivredi, životnoj sredini i javnom zdravlju. Učesnica je na svetskim izazovima iz nauke o podacima i sa timom istraživača postigla je značajne rezultate: treće mesto na Nokia Mobile Data Challenge (2012), finalista Orange France, Data for Development Challenge (2013), finalista / pobednik / treće mesto na Singenta Crop Challenge (2016, 2017, 2019) i finalista UNDP izazova depopulacije Srbije (2020).

&nbsp;

Predrag Lugonja je istraživac na Institutu BioSens. Diplomirao je na Departmanu za Energetiku, elektroniku i telekomunikacije Fakulteta tehničkih nauka Univerziteta u Novom Sadu. Student je završne godine doktorksih studia na Fakulteta tehničkih nauka Univerziteta u Novom Sad. Trenutno se bavi istraživanjem na poljima GIS i podataka koje su dobijeni daljinskom detekcijom.

Dr Oskar Marko je rukovodilac Centra za informacione tehnologije i pomoćnik direktora za inovacije i saradnju sa privredom. Njegovo istraživanje usmereno je ka primeni naprednih metoda mašinskog učenja i evolutivnih algoritama u poljoprivredi. Treću godinu osnovnih studija je pohađao je na “Siti univerzitetu” u Londonu, gde je odbranio diplomski rad u domenu obrade signala. Vodio je tim BioSensa koji je razvio nove algoritme iz oblasti obrade velikih podataka koje su za cilj imale predikciju prinosa, distribuciju semena i optimizaciju strategije setve, koje su Institutu osigurale prvo mesto na Sindžentinom “Izazovu za useve” i CGIAR-ovom “Inspirišućem izazovu”. Aktivno je uključen u mnoge projekte iz poziva Horizont 2020, među kojima su Antares, Cybele, Dragon i Flexirobots, kao i u projekte sa kompanijama poput Krivaje, MK Agri, Delte i drugih poljoprivrednih kompanija i osiguravajućih kuća.

Miloš Pandžić je istraživač pripravnik čije interesovanje pokriva nekoliko istaživačkih oblasti: osmatranje Zemlje, daljinsku detekciju, obradu slike i analitiku podataka. Trenutno je student doktorskih studija odseka za Geodeziju i geoinformatiku na Građevinskom fakultetu, Univerziteta u Beogradu, gde je takođe završio i osnovne i master akademske studije. U 2015. godini dobio je stipendiju vlade Austrije (OeAD) i proveo 4 meseca na Tehničkom univerzitetu u Beču gde je analizirao geometrijski kvalitet Sentinel-2 satelitskih multispektralnih snimaka. Iz ovog istraživanja je proizašao i Milošev master rad. Nakon završetka studija, radio je u Finskom geoprostornom istraživačkom institutu (Finnish Geospatial Research Institute - FGI) na više projekata finansiranih od strane Finske Akademije i Evropske svemirske agencije. Tu je radio na segmentaciji i detekciji promena na satelitskim slikama za potrebe poljoprivrede, urbanizma i šumarstva. U 2018. godini pridružio se Institutu BioSens gde radi na fuziji podataka optičkih i radarskih slika za procenu prinosa, klasifikaciju useva, praćenje razvoja useva, itd. Učestvuje u realizaciji nekoliko nacionalnih i internacionalnih projekata uključujući Horizon 2020 projekte Antares, Cybele, Dragon, PARSEC i agROBOfood.

Dr Marko Panić je naučni saradnik u oblasti tehnološko-tehničkih nauka – informacionih tehnologija na institutu BioSens. Tema doktorata mu je vezana za razvoj algoritama za ubrzanu rekonstrukciju slika iz nepotunih merenja korišćenjem retke reprezentacije signala I statističkih Markovljevih slučajnih modela Njegov pravac istraživanja obuhvata statističko modelovanje multidimenzionalnih podataka dobijenih sa različitih senzora. Trenutno radi na analizi I obradi slika dobijenih sa hiperspektralnih I multispektralnih kamera korišćenjenem naprednih metoda mašinskog učenja sa primenom u proceni kvaliteta voća I povrća. Kao član tima BioSens ucestvovao je na Sindžentinom takmičenju “Izazovu za useve” koje rezltiralo osvajanjem prvog mesta. Aktivno je uključen u mnoge projekte iz poziva Horizont 2020, među kojima su Antares, Cybele, Dragon i Flexirobots, kao i u projekte sa kompanijama poput Krivaje, MK Agri i Delta Agrar.

Dejan Pavlović je istraživač na Institutu Biosens i student doktorskih studija na Elektrotehničkom fakultetu, Univerziteta u Beogradu. Njegovo primarno interesovanje obuhvata obradu signala i primenu mašinskog učenja u poljoprivredi. Dejan je stekao zvanje diplomiranog inženjera biomedicinskog inženjerstva na Fakultetu tehničkih nauka, Univerziteta u Novom Sadu i zvanje master inženjera elektrotehnike i računarstva na Elektrotehničkom fakultetu, Univerziteta u Beogradu. Trenutno radi na razvoju algoritma za klasifikaciju različitih aktivnosti kod krava primenom Konvolucionih neuralnih mreža. Učestvuje u realizaciji nekoliko nacionalnih i internacionalnih projekata uključujući Horizon 2020 projekte, Antares i Dragon.

Branislav Pejak je student doktorskih studija Elektrotehničkog i kompjuterskog inženjerstva na Fakultetu tehničkih nauka, Univerziteta Novi Sad, na kojem je takođe završio osnovne (2017) i master (2018) studije u oblsti Biomedicinskog inženjerstva. Tokom osnovnih i master studija bio je stipendista Ministarstva obrazovanja, nauke i tehnološkog razvoja, Republike Srbije. Od marta 2018. godine zaposlen je u Centru za informacione tehnologije, Instituta BioSens, na poziciji mlađeg istraživača. Tema njegovog istraživanja je analiza podataka sa satelitskih snimaka. Trenutno se bavi razvojem GAN algoritama mašinskog učenja za klasifikaciju i predikciju prinosa na bazi multispektralnih satelitskih snimaka. Aktivno je uključen u mnoge projekte sa poziva Horizont 2020, među kojima su Antares, Cybele, Dragon i AgRoboFood koji imaju za cilj upotrebu informaciono-komunikacionih tehnologija u poljoprivredi poput kreiranja mapa za selektivno uzorkovanje i đubrenje zemljišta, detekciju navodnjavanih površina, procenu prinosa i proteinskog sastava useva.

&nbsp;

Bojana Ivošević se bavi upotrebom bespilotnih letelica za efikasno nadgledanje ratarskih kultura, slatkih voda i životne sredine. Ona je po struci biolog i ekolog. Doktorske studije je upisala na Kyungpook Nacionalnom Univerzitetu u Južnoj Koreji, gde je stekla dve godine iskustva rada na bespilotnim letelicama. Nastavila je studije na Prirodno-matematičkom fakultetu, Univerzitetu u Novom Sadu. Njen trenutni fokus istraživanja je primena bespilotnih letelica u preciznoj poljoprivredi za izračunavanje vegetacijskih indeksa važnih za procenu prinosa, detekciju bolesti i upravljanje usevima. Takođe je zainteresovana za primenu dronova u detekciji i prebrojavanju životinjskih vrsta, praćenju cvetanja algi, klasifikaciju zemljišnog pokrivača i sezonskih efekata na ekosisteme.

 Ovaj projekat je finansiran od strane Evropske Unije, iz programa za Istraživanje i Inovacije-           Horizont 2020, pod sporazumom o grantu broj 101017111. 


Horizont 2020

Za analizu problema i prikupljanje podataka, BioSens koristi različite uredjaje za akviziciju slike. Razlike između uređaja tiču se širine spektra elektromagnetnog zračenja i broja spektralnih pojasa u kojima rade. Hiperspektralne kamere obuhvataju vidljivi i blisko-infracrveni deo spektra (VNIR) kao i isključivo blisko-infracrveni deo spektra (NIR) i koriste se za laboratorijske analize, dok se termalna i multispektralna kamera uglavnom koriste za terenski rad.

&nbsp;

Laboratorijske analize su preciznije jer hiperspektralne kamere mogu imati rezoluciju i od preko 200 kanala, a nakon kalibracije korišćenjem referentih ploča se dodatno povećava pouzdanost merenja. Ipak, kako jedna hiperspektralna slika može imati i do 2GB, ovakve slike zahtevaju velik memorijski prostor i veoma efikasne algoritme obrade slike. Za redukciju količine podataka biraju se najinformativniji spektralni kanali sa visokom odnosom snage signala i šuma ili se primenjuju standardni postupci analize glavnih komponenata (PCA ili ICA). Slike koji stižu sa termalnih i multispektralnih kamera takođe zahtevaju predobradu ali su ovi postupci više fokusirani na smanjenje šuma filtriranjem slika. Sledeći korak, zavisno od problema koji se razmatra, je analiza slike koja obuhvata tehnike izdvajanja obeležja, grupisanja podataka po sličnosti i primene dubokih neuralnih mreža radi segmentovanja slike na regione od interesa. Dalje se koriste dodatni algoritmi za finalnu prezentaciju rezultata i dobijeni rezultati se prosleđuju algoritmima mašinskog učenja za dalju analizu podataka.

&nbsp;

<img class="alignnone wp-image-1839 size-large" src="https://api.biosens.rs/wp-content/uploads/2021/06/100-1024x544.jpg" alt="" width="1024" height="544" />

&nbsp;

U praksi se ove napredne kamere koriste za razvoj algoritama za detekciju infekcija na povrću ili otkrivanja modrica na voću i to nekoliko dana pre nego što simptomi postanu vidljivi golim okom. Značajan je i njihov doprinos u praćenju uslova ekosistema i predikciji prinosa, a razvijeni algoritmi pružaju dodatnu informaciju o stanju biljaka, daju procenu roka trajanja poljoprivrednih proizvoda kao, predviđaju prinos i daju objektivne predloge za donošenje odluka nakon žetve.

Hiperspektralne, multispektralne i termalne slike

Upotreba robotike u poljoprivredi je oblast u nauci koja se sve više razvija i sa novim rešenjima dospeva na tržište. Značajan napredak u ovoj oblasti uzrokovan je nedostatkom radne snage, porastom troškova proizvodnje kao i zamorom i repetitivnošću koje određeni zadaci nose sa sobom. BioSens koristi robote kao platforme koji se kreću i prikupljaju informacije o usevu. Ove platforme su opremljene senzorima koji su komercijalno dostupni i senzorima koje je razvila grupa za dizajn senzora Instituta BioSens.

&nbsp;

Najveći projekat na kom je institut radio je vezan za analize tla na terenu. Predloženi sistem se kreće kroz polje i vrši uzorkovanje zemljišta i analizu parametara (makronutrijenti, mikronutrijenti, pH, elektroprovodljivost) kao i georeferenciranje rezultata analize. Kompletan proces je autonoman jer je moguće unapred definisati rutu koju će robotska platforma preći na parceli koristeći GPS koordinate i 3D Real-Time-LIDAR sistem za rekonstrukciju okruženja i izbegavanje mogućih prepreka. Kao rezultat analize, generišu se višeslojne mape gde svaki sloj predstavlja jedan od parametara dobijenih predloženim sistemom. Poljoprivrednik će pomoću ovih mapa moći da donese pravu odluku, poboljša proizvodnju i u potpunosti iskoristi preciznu poljoprivredu u pogledu efikasnosti, povećanja prinosa i održivosti poljoprivredne proizvodnje.

&nbsp;

&nbsp;

<img class="alignnone wp-image-1493 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/51.jpg" alt="" width="672" height="422" />

Robotika

Na Institutu BioSens koristimo slike kvadkoptera DJI Inspire 1 kako bismo dobili bolji uvid u stanje useva i preciznije planirali i upravljali agroekološkim aktivnostima. Ova tehnologija omogućava prikupljanje podataka visoke rezolucije ispod nivoa oblaka sa mnogo više detalja nego što se može videti u satelitskim snimcima. Skup podataka uključuje slike iz vidljivog dela spektra kao i slike dobijene modifikovanom NDVI kamerom. Takođe koristimo i Flir termalnu kameru za merenje temperature na velikim površinama u hetoregenim staništima. Bespilotne letelice su jednostavne za upotrebu, jer se većina mapiranja terena i prikupljanja podataka sada sprovodi autonomno, prema zadatim granicama terena uz pomoć aplikacija za planiranje leta.

&nbsp;

Za preklapanje slika koristimo softvere Pix4D i AgisoftPhotoscan kako bismo generisali ortomozaike, mape vegetacionih indeksa i digitalne modele površine. Ove orto-mape su važni slojevi informacija i mogu se integrisati u GIS sisteme zajedno sa slojevima podataka dobijenih iz drugih izvora. Ovakvi skupovi podataka se zatim obrađuju i koriste za obuku naprednih algoritama mašinskog učenja.

&nbsp;

BioSens koristi bespilotne vazdušne sisteme za otkrivanje nedostatka hranljivih materija u biljkama, napada štetočina, prekomerne suše ili vlage kao i drugih faktora koji utiču na razvoj useva, kao i za povećanje efikasnosti pri donošenju odluka. Posebno su korisni za površine od nekoliko desetina hektara, poput voćnjaka i vinograda, jer skeniranje velikih ratarskih parcela zahteva mnogo vremena. Bespilotne letelice su se pokazale i kao efikasni alati za praćenje stanja ekosistema. Pokrivajući veća prostranstva, omogućavaju pristup informacijama o tipu staništa u kojem rastu i razvijaju se živi organizmi od insekata do sisara. Njihova upotreba dovela je do revolucionarnih otkrića na poljima prostorne ekologije i poljoprivrede, i omogućila nam dobijanje podataka visoke rezolucije za donošenje optimalnih odluka sa visokom preciznošću.

&nbsp;

&nbsp;

<img class="alignnone wp-image-1853 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/slika5.png" alt="" width="775" height="444" />

NDVI map of sugar beet crop

Obrada slika sa dronova

BioSens je dom habu za Sentinel podatke i kao takav je korisnik proizvoda Kopernikus programa Evropske svemirske agencije (ESA). Pre lansiranja Sentinel satelita, naša istraživanja su se zasnivala na snimcima sa satelita Landsat 7 i 8, kao i RapidView-a, ali su danas primarno usmerena na podatke koji potiču sa Sentinel 1 i 2 misija. Slike sa Sentinela 2 sadrže 13 spektralnih kanala i dostupne su na svakih 5 dana u zavisnosti od oblačnosti, dok Sentinel 1 koji snima radarske slike omogućava praćenje čak i u slučaju oblačnosti. Na ovaj način omogućeno je kontinualno osmatranje površine Zemlje u visokoj rezoluciji uključujući praćenje poljoprivrednje proizvodnje kroz sve faze rasta biljaka.

&nbsp;

&nbsp;

<img class="alignnone wp-image-1495 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/52.png" alt="" width="960" height="600" />

&nbsp;

Od 2012. godine proizvodimo mape upotrebe zemljišta za Autonomnu Pokrajinu Vojvodinu, gde klasifikujemo zemljište u šume, vodene mase, urbane površine i različite tipove useva. Ovakve mape su nam omogućile da mapiramo životinjska staništa, posmatramo kako se menjaju tokom godina i ispratimo plodored na proizvodnim parcelama. Oslanjajući se na principe fuzije podataka, uspeli smo da dođemo do sub-pikselske rezolucije korišćenjem višestrukih izvora podataka sa heterogenim mrežama piksela. Druga grana našeg istraživanja se bavi prognozom prinosa useva. Na osnovu podataka sa terena dobijenih pomoću digitalnih knjiga polja, kraudsorsinga ili mapa prinosa, obučavamo napredne modele mašinskog učenja u cilju što bolje procene prinosa. Ovakvi modeli koriste više od 30 satelitskih vegetacionih indeksa koji uključuju domensko znanje o usevima, a njihova tačnost se povećava kako napredujemo kroz faze razvoja biljaka. Vegetacioni indeksi su takođe odlični pokazatelji procene štete jer neki od njih veoma dobro reaguju na pojavu prirodnih nepogoda poput suše, požara i izmrzavanja. Na ovaj način se osiguravajuća društva uključuju u partnerstvo sa Institutom, zajedno sa državnim oragnima, istraživačkim institucijama i poljoprivrednim proizvođačima koji naše rezultate koriste za optimalno donošenje odluka o navodnjavanu i distribuciji đubriva i pesticida.

Obrada satelitskih slika

Analitika velikih podataka se zasniva na obradi heterogenih izvora podataka, različitih rezolucija, koji se prikupljaju u nejednakim vremenskim intervalima. Osim satelitskih podataka, Grupa za veštačku inteligenciju Instituta BioSens prikuplja i koristi podatke dobijene pomoću elektromagnetne sonde koja meri električnu provodnost zemljišta, kao i brojne druge instrumente koji se montiraju na poljoprivredne mašine ili koriste iz ruke. Električna provodnost zemljišta je odličan indikator teksture zemljišta, njene kompaktnosti i hemijskog sastava. Ovi parametri mogu se odrediti na različitim dubinama u zavisnosti od potreba i interesovanja korisnika prema tipu biljaka. Mape elevacije terena nisu ništa manje važne s obziorm na to da određuju pravac i smer oticanja nutrijenata koji je u tesnoj sprezi sa prinosom, dok se ručni uređaji, poput Plant-O-Meter-a razvijenog od strane istraživača Instituta BioSens, uspešno koriste pri monitoringu rasta useva.

&nbsp;

<img class="alignnone wp-image-1486 " src="https://api.biosens.rs/wp-content/uploads/2021/06/46.jpg" alt="" width="220" height="312" /><img class="alignnone wp-image-2435" src="https://api.biosens.rs/wp-content/uploads/2021/06/prostorna-212x300.jpg" alt="" width="220" height="311" />

<img class="alignnone wp-image-1487" src="https://api.biosens.rs/wp-content/uploads/2021/06/47-212x300.jpg" alt="" width="220" height="311" /><img class="alignnone wp-image-1490" src="https://api.biosens.rs/wp-content/uploads/2021/06/9-212x300.jpg" alt="" width="220" height="311" />

&nbsp;

Korišćenjem tehnika fuzije podataka, heterogeni podaci se agregiraju u jedinstven skup koji se koristi za iscrtavanje, odnosno određivanje menadžment zona. Određivanje zona vrši se algoritmima klasterizacije bilo u normalnoj ili težinskoj formi, što zavisi od relativne važnosti ulaznih podataka. Ove menadžment zone međusobno imaju različite osobine, ali se u okviru jedne zone zemljište može posmatrati homogenim. To dalje omogućava davanje preporuka o poljoprivrednoj proizvodnji za svaku zonu ponaosob, u skladu sa njenim specifičnostima. Ovakav pristup donošenju odluka u okviru jedne parcele osnova je za preciznu poljoprivredu, a ujedno je i prvi korak pri određivanju optimalne lokacije za postavljanje senzora jer se na taj način maksimizuje njihova efikasnost.

Analitika prostornih podataka

Sa pojavom neparametaskih modela mašinskog učenja i rastućom popularnošću dubokih neuronskih mreža, većina klasifikatora i regresora koji se danas koriste su “crne kutije”. Tačnost ovih modela često je neuporedivo bolja nego kod konvencionalih modela, a iako znamo matematiku koja se krije iza njih, ipak ne možemo direktno da zaključimo kompleksnu vezu izmeću ulaznih i izlaznih podataka. Velika tačnost modela je od izuzetnog praktičkog značaja, ali kako bismo imali bolje razumevanje problema i kako bismo pomerili granice nauke moramo da zagrebemo dublje u način njihovog funkcionisanja.

&nbsp;

<img class="alignnone wp-image-1474 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/39-1.jpg" alt="" width="1187" height="1028" />

&nbsp;

&nbsp;

Oslanjamo se na nekoliko tehnika otkrivanja znanja, koje se baziraju na indirektnim posmatranjima modela mašinskog učenja. Jedan od njih je test slučajne permutacije. Pomoću njega procenjujemo važnost obeležja na osnovu smanjenja tačnosti modela nakon slučajne permutacije obeležja u skupu podataka. Morisonova i Soboljeova analiza neizvesnosti i osetljivosti nam sa druge strane govore o mogućim izlazima sistema i interakcijama između ulaznih parametara. Analiza objašnjivosti modela ide korak dalje. Za svako obeležje koje se koristi u prediktivnom modelu, kroz aproksimaciju zasnovanu na uzorkovanju , možemo da procenimo pozitivan ili negativan uticaj na izlaznu promenljivu, što nam omogućava da domensko istraživanje usmerimo tamo gde je najefikasnije. Ovi i drugi modeli nam omogućavaju da odredimo uticaj različitih parametara zemljišta i vremena na prinos useva i da utvrdimo koji hibridi kukuruza su otporni na sušu, a koje sorte soje imaju stabilan prinos.

&nbsp;

&nbsp;

<img class="alignnone wp-image-1471 " src="https://api.biosens.rs/wp-content/uploads/2021/06/38-1024x768.jpg" alt="" width="351" height="263" /><img class="alignnone wp-image-1473 " src="https://api.biosens.rs/wp-content/uploads/2021/06/39-1024x768.jpg" alt="" width="351" height="263" />

Otkrivanje znanja

Značajan deo BioSensovih istraživanja u oblasti Velikih Podataka je fokusiran na dva tipa podataka koji zahtevaju velik memorijski kapacitet za skladištenje i ekstremno efikasne algoritme za obradu podataka koji mogu da se izvrše u zahtevanom vremenskom roku.

&nbsp;

Prvi od njih su satelitske slike. Sentinel 2 sateliti isporučuju nove slike u 13 pojasa za bilo koju tačku na Zemlji, približno svakih 5 dana. Ove slike dolaze u isečcima dimenzija 100x100 km2, sa prostornom rezolucijom od 10m. U okviru Cybele Projekta, na primer, analiziramo satelitske slike cele Evrope da bismo klasifikovali polja soje, što znači da treba da prođemo kroz stotine isečaka i primenimo nad njima algoritam za klasifikaciju. U takvim situacijama, radimo paralelizaciju koda radi bolje i efikasnije eksploatacije multicore server arhitekture, ubrzavajući procese korišćenjem biblioteke Cython, dok u isto vreme prilagođavamo implementaciju da umanjimo računarsku kompleksnost algoritama.

&nbsp;

Drugi tip podataka su liste poziva korisnika mobilne telefonije i oni zadovoljavaju sve “V”-ove analitike velikih podataka. Ovi podaci se sastoje iz liste uspostavljenih poziva, razmenjenih poruka i registrovanog internet saobraćaja i ogromne su veličine. Važno pitanje u obradi je vezano za privatnost i izdvajanje korisnih informacija iz anonimizovanih i agregiranih skupova podataka predstavlja veliki izazov. Takođe, iako su podaci semi-struktuirani, tradicionalna RDBMS tehnologija za analitiku podataka nije primenljiva na skupove ove veličine. U tu svrhu razvijamo tokove rada za obradu korišćenjem Apache porodice tehnologija - Spark, Hadoop i Hive. One nam daju mogućnost da dobijemo dodatni uvid u ljudsko ponašanje, šablone kretanja i saznamo više o interakciji između ruralnih i urbanih oblasti.

&nbsp;

<img class="alignnone wp-image-1465 size-large" src="https://api.biosens.rs/wp-content/uploads/2021/06/35-1024x578.jpg" alt="" width="1024" height="578" />

Analitika Velikih Podataka

Duboko učenje je novi metod mašinskog učenja. Zbog svoje specifičnosti i odličnih performansi u obradi slike često se posmatra kao zasebna oblast u okviru veštačke inteligencije. Najbolje radi na velikim skupovima podataka koji imaju multi-dimenzionalne karakteristike, sa kakvim se često susrećemo na BioSensu.

&nbsp;

U prvu grupu podataka koje posedujemo spadaju hiperspektralne, multispektralne i termalne slike, napravljene u laboratoriji ili na terenu. Cilj njihove obrade je segmentacija slike i prepoznavanje karakterističnih obrazaca, kao što su počeci truljenja i fizička oštećenja na jabukama, infekcija peteljke paradajza ili detektovanje klasova pšenice. Brojne arhitekture su testirane sa različitim parametrima, poput broja konvolucionih i potpuno povezanih slojeva, vrsta aktivacionih funkcija i slučajne eliminacije dela mreže da bi se izbeglo preučenje. Svaka arhitektura je optimizovana za specifičan problem, uzimajući u obzir veličinu slika i broj uzoraka u bazi podataka.

&nbsp;

<img class="alignnone wp-image-1469 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/337.jpg" alt="" width="640" height="480" />

&nbsp;

Druga grupa skupa podataka se sastoji od merenja čestica u vazduhu pomoću uređaja Rapid-E, prvog automatskog detektora čestica zasnovanog na tzv. LIF (eng. laser-induced ﬂuorescence) tehnologiji. LIF tehnologija beleži raspršene fluroscentne obrasce svetlosti i lasera, koji predstavljaju oblik i hemijski otisak čestica. U ovom slučaju se koristi jedan od primera dubokih neuronskih mreža (DNN) koje se najviše oslanjaju na konvolucione slojeve, zbog njihove sposobnosti da obrade skupove podataka koji potiču iz različitih izvora.

&nbsp;

U oba slučaja, predobrada je jedan od najvažnijih koraka, jer su zašumljeni i nestandardni uzorci veoma teški za analizu. Takođe, važno je istaći da svaki problem zahteva drugačije, prilagođeno rešenje, čija složenost odgovara veličini i složenosti skupova podataka i čija tačnost mora opravdati vreme učenja i izvršavanja algoritma.

&nbsp;

Duboko učenje

Cilj BioSensove Grupe za veštačku inteligenciju je da obezbedi sveobuhvatan i precizan uvid u stanje biosistema, poveća efikasnost fenotipizacije i oplemenjivanja biljaka, ponudi bolje razumevanje molekularne biologije i pomogne poljoprivrednicima kroz optimalno i automatsko donešenje odluka. Naše istraživanje počinje od problema i zahteva inovativnu primenu i razvoj algoritama koji su poslednja reč nauke. Algoritmi mašinskog učenja koji se najčešće koriste za probleme kojima se BioSens bavi su stabla odluke, vektori nosači, ridž regresija, spektralno klasterovanje i mnogi drugi, uključujući i posebnu porodicu algoritama mašinskog učenja – duboko učenje, a tamo gde složenost problema prevazilazi postojeće algoritme, istraživači sa Instituta razvijaju nove.

&nbsp;

Primer algoritma koji je razvijen na BioSens Institutu je algoritam regresije težinskim histogramima. Dati algoritam je zasnovan na mehanizmu glasanja, gde uzorci za obuku glasaju za izlaznu vrednost test uzorka. Svaki glas dobija određenu težinu u zavisnosti od sličnosti uzorka koji glasa i uzorka iz test skupa. Pomoću dobijenih glasova, formira se histogram koji de fakto predstavlja funkciju gustine verovatnoće (PDF) izlaza test uzorka, dok matematičko očekivanje PDF-a predstavlja predikovanu vrednost datog uzorka. Ovaj algoritam se pokazao veoma korisnim za problem predikcije prinosa, gde je nadmašio algoritme stabla odluke, vektora nosača i veštačkih neuronskih mreža.

&nbsp;

&nbsp;

<img class="alignnone wp-image-1478 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/42.png" alt="" width="503" height="504" />

&nbsp;

&nbsp;

Još jedan inovativni algoritam koji je razvijen na Institutu BioSens je algoritam integrisane klasterizacije zasnovan na nenegativnoj faktorizaciji matrica. Kombinovanjem individualnih rezultata klasterovanja u ansambl, dokazi se akumuliraju u cilju poboljšanja rezultata klasterizacije.

Individualni rezultati klasteriziacije spojeni su u zajedničku matricu koja se zatim faktorizuje u matrice kodovanja i bazne vektore. Algoritam ima mogućnost da integriše klastere koji potiču iz različitih baza podataka, različitih nivoa predprocesiranja ili poseduju različite podskupove obeležja.
Ovaj algoritam se pokazao kao izuzetno koristan za funkcionalnu genomiku. Dati pristup, zanosvan na nenegativnoj faktorizaciji matrica poseduje mogućnost fuzije različitih izvora podataka i rezultuje klasterima gena sa povećanom funkcionalnošću i pokrivenošću gena.

&nbsp;

&nbsp;

<img class="alignnone wp-image-1480 size-full" src="https://api.biosens.rs/wp-content/uploads/2021/06/43.png" alt="" width="1691" height="692" />

Mašinsko učenje

Mašinsko učenje i domenska ekspertiza se često posmatraju kao dva rivalska pristupa u rešavanju problema u raznim oblastima, uključujući poljoprivredu. Međutim, naša strategija je da se koristi najbolje iz oba pristupa. Algoritmi mašinskog učenja daju odlične rezultate pri otkrivanju zavisnosti među podacima, ali uveliko postoje i dobro poznate teorije koje objašnjavaju prirodne procese i koje su već dokazane u praksi. Mi u potpunosti iskorištavamo ovu činjenicu ugrađujući domensko znanje u naše modele. To se obično radi tokom procesa izbora i inženjeringa obeležja. Dobro osmišljena obeležja uvode novi nivo informacija u sistem i usmeravaju algoritme mašinskog učenja ka optimalnim konfiguracijama.

&nbsp;

Jedan primer inženjeringa obeležja, zasnovan na domenskoj ekspertizi, je korišćen pri izboru hibrida kukuruza otpornih na stres u Syngenta izazovu za useve u 2019. godini. Suša je pojava povezana sa dužim periodom vremena bez padavina, praćenog visokim temperaturama i niskom vlažnošću vazduha. Međutim, ne postoji stroga matematička definicija suše i u praksi se koriste razni pokazatelji. Na osnovu dnevnih vrednosti maksimalne i minimalne temperature vazduha, količine padavina i pritiska vodene pare u nekoliko faza tokom vegetacione sezone, dizajnirali smo obeležja koja ukazuju na pojavu suše ili toplotnog stresa na određenoj lokaciji. Među njima su bili maksimalan broj uzastopnih suvih dana, mala količina padavina u određenoj fazi razvoja i deficit vodene pare. U ovom problemu su nam dizajnirana obeležja pomogla da pronađemo hibride otporne na sušu koje treba posejati kako bi se maksimizovao prinos i minimizovao rizik loše žetve, ali se ovaj pristup može primeniti i na druge probleme. Ostali primeri sadrže dizajn meteoroloških obeležja za predikciju koncentracije polena u vazduhu, dizajn elevacionih obeležja za bolju predikciju slivanja nutrijenata na njivama i dizajn obeležja koja opisuju kretanje životinja u predikciji bolesti i inflamacije mlečnih žlezda.

&nbsp;

<img class="alignnone wp-image-1471 size-large" src="https://api.biosens.rs/wp-content/uploads/2021/06/38-1024x768.jpg" alt="" width="1024" height="768" />

<img class="alignnone wp-image-1473 size-large" src="https://api.biosens.rs/wp-content/uploads/2021/06/39-1024x768.jpg" alt="" width="1024" height="768" />

FLEXIGROBOTS-Flexible robots for intelligent automation of precision agriculture operations

Themes:

People:

Centers:

CST

CIRP

CIT