Gergely Teschner
58030586300
Publications - 10
Aerial Image-Based Crop Row Detection and Weed Pressure Mapping Method
Publication Name: Agronomy
Publication Date: 2025-08-01
Volume: 15
Issue: 8
Page Range: Unknown
Description:
Accurate crop row detection is crucial for determining weed pressure (weeds item per square meter). However, this task is complicated by the similarity between crops and weeds, the presence of missing plants within rows, and the varying growth stages of both. Our hypothesis was that in drone imagery captured at altitudes of 20–30 m—where individual plant details are not discernible—weed presence among crops can be statistically detected, allowing for the generation of a weed distribution map. This study proposes a computer vision detection method using images captured by unmanned aerial vehicles (UAVs) consisting of six main phases. The method was tested on 208 images. The algorithm performs well under normal conditions; however, when the weed density is too high, it fails to detect the row direction properly and begins processing misleading data. To investigate these cases, 120 artificial datasets were created with varying parameters, and the scenarios were analyzed. It was found that a rate variable—in-row concentration ratio (IRCR)—can be used to determine whether the result is valid (usable) or invalid (to be discarded). The F1 score is a metric combining precision and recall using a harmonic mean, where “1” indicates that precision and recall are equally weighted, i.e., β = 1 in the general Fβ formula. In the case of moderate weed infestation, where 678 crop plants and 600 weeds were present, the algorithm achieved an F1 score of 86.32% in plant classification, even with a 4% row disturbance level. Furthermore, IRCR also indicates the level of weed pressure in the area. The correlation between the ground truth weed-to-crop ratio and the weed/crop classification rate produced by the algorithm is 98–99%. As a result, the algorithm is capable of filtering out heavily infested areas that require full weed control and capable of generating weed density maps on other cases to support precision weed management.
Open Access: Yes
Predicting maize growth and biomass: Integrating gradient boosted trees with sentinel images and IoT
Publication Name: Progress in Agricultural Engineering Sciences
Publication Date: 2025-01-01
Volume: Unknown
Issue: Unknown
Page Range: Unknown
Description:
Agricultural big data and high-performance computing have significantly improved crop yield modeling. Maize growth dynamics and yield prediction are crucial for sustainable agriculture. This study introduces an advanced modeling approach utilizing Gradient Boosted Decision Trees (GBDT) combined with a feature selection strategy to predict maize biomass production. A dataset of 200 unique maize plants was observed throughout the vegetation season. Our approach integrates manual measurements, meteorological data, and vegetation indices along with Internet of Things (IoT) field sensors to perform spatio-temporal analysis. Results indicate that maize stalk thickness and height are the most reliable predictors of biomass yield, while environmental variables show minimal impact. The most effective model, period-dependent GBDT, demonstrated superior predictive performance, achieving an average error of 4.39 mm in plant growth predictions. Notably, stalk thickness and height can be estimated six weeks before harvest, while biomass yield two weeks before harvest. This research underscores the potential of machine learning and remote sensing to enhance precision agriculture decision-making.
Open Access: Yes
Field-grown tomato yield estimation using point cloud segmentation with 3D shaping and RGB pictures from a field robot and digital single lens reflex cameras
Publication Name: Heliyon
Publication Date: 2024-10-30
Volume: 10
Issue: 20
Page Range: Unknown
Description:
The aim of this study was to estimate field-grown tomato yield (weight) and quantity of tomatoes using a self-developed robot and digital single lens reflex (DSLR) camera pictures. The authors suggest a new approach to predicting tomato yield that is based on images taken in the field, 3D scanning, and shape. Field pictures were used for tomato segmentation to determine the ripeness of the crop. A convolution neural network (CNN) model using TensorFlow library was devised for the segmentation of tomato berries along with a small robot, which had a 59.3 % F1 score. To enhance the accurate tomato crop model and to estimate the yield later, point cloud imaging was applied using a Ciclops 3D scanner. The best fitting sphere model was generated using the 3D model. The most optimal model was the 3D model, which gave the best representation and provided the weight of the tomatoes with a relative error of 21.90 % and a standard deviation of 17.9665 %. The results indicate a consistent object-based classification of the tomato crop above the plant/row level with an accuracy of 55.33 %, which is better than in-row sampling (images taken by the robot). By comparing the measured and estimated yield, the average difference for DSLR camera images was more favorable at 3.42 kg.
Open Access: Yes
Weed detection in agricultural fields using machine vision
Publication Name: Bio Web of Conferences
Publication Date: 2024-08-23
Volume: 125
Issue: Unknown
Page Range: Unknown
Description:
Weeds have the potential to cause significant damage to agricultural fields, so the development of weed detection and automatic weed control in these areas is very important. Weed detection based on RGB images allows more efficient management of crop fields, reducing production costs and increasing yields. Conventional weed control methods can often be time-consuming and costly. It can also cause environmental damage through overuse of chemicals. Automated weed detection and control technologies enable precision agriculture, where weeds are accurately identified and targeted, minimizing chemical use and environmental impact. Overall, weed detection and automated weed control represent a significant step forward in agriculture, helping farmers to reduce production costs, increase crop safety, and develop more sustainable agricultural practices. Thanks to technological advances, we can expect more efficient and environmentally friendly solutions for weed control in the future. Developing weed detection and automated control technologies is crucial for enhancing agricultural efficiency. Employing RGB images for weed identification not only lowers production costs but also mitigates environmental damage caused by excessive chemical use. This study explores automated weed detection systems, emphasizing their role in precision agriculture, which ensures minimal chemical use while maximizing crop safety and sustainability.
Open Access: Yes
Weed Detection and Classification with Computer Vision Using a Limited Image Dataset
Publication Name: Applied Sciences Switzerland
Publication Date: 2024-06-01
Volume: 14
Issue: 11
Page Range: Unknown
Description:
In agriculture, as precision farming increasingly employs robots to monitor crops, the use of weeding and harvesting robots is expanding the need for computer vision. Currently, most researchers and companies address these computer vision tasks with CNN-based deep learning. This technology requires large datasets of plant and weed images labeled by experts, as well as substantial computational resources. However, traditional feature-based approaches to computer vision can extract meaningful parameters and achieve comparably good classification results with only a tenth of the dataset size. This study presents these methods and seeks to determine the minimum number of training images required to achieve reliable classification. We tested the classification results with 5, 10, 20, 40, 80, and 160 images per weed type in a four-class classification system. We extracted shape features, distance transformation features, color histograms, and texture features. Each type of feature was tested individually and in various combinations to determine the best results. Using six types of classifiers, we achieved a 94.56% recall rate with 160 images per weed. Better results were obtained with more training images and a greater variety of features.
Open Access: Yes
DOI: 10.3390/app14114839
Yield Prediction Using NDVI Values from GreenSeeker and MicaSense Cameras at Different Stages of Winter Wheat Phenology
Publication Name: Drones
Publication Date: 2024-03-01
Volume: 8
Issue: 3
Page Range: Unknown
Description:
This work aims to compare and statistically analyze Normalized Difference Vegetation Index (NDVI) values provided by GreenSeeker handheld crop sensor measurements and calculate NDVI values derived from the MicaSense RedEdge-MX Dual Camera, to predict in-season winter wheat (Triticum aestivum L.) yield, improving a yield prediction model with cumulative growing degree days (CGDD) and days from sowing (DFS) data. The study area was located in Mosonmagyaróvár, Hungary. A small-scale field trial in winter wheat was constructed as a randomized block design including Environmental: N-135.3, P2 O5 -77.5, K2 O-0; Balance: N-135.1, P2 O5 -91, K2 O-0; Genezis: N-135, P2 O5 -75, K2 O-45; and Control: N, P, K 0 kg/ha. The crop growth was monitored every second week between April and June 2022 and 2023, respectively. NDVI measurements recorded by GreenSeeker were taken at three pre-defined GPS points for each plot; NDVI values based on the MicaSense camera Red and NIR bands were calculated for the same points. Results showed a significant difference (p ≤ 0.05) between the Control and treated areas by GreenSeeker measurements and Micasense-based calculated NDVI values throughout the growing season, except for the heading stage. At the heading stage, significant differences could be measured by GreenSeeker. However, remotely sensed images did not show significant differences between the treated and Control parcels. Nevertheless, both sensors were found suitable for yield prediction, and 226 DAS was the most appropriate date for predicting winter wheat’s yield in treated plots based on NDVI values and meteorological data.
Open Access: Yes
The Effect of Illumination on HSV Colour Segmentation for Ripe Tomatoes based on Machine Vision
Publication Name: Chemical Engineering Transactions
Publication Date: 2024-01-01
Volume: 114
Issue: Unknown
Page Range: 829-834
Description:
In agriculture, computer vision and image processing are essential for monitoring crops and controlling robots and actuators. In this work, the detection of ripe tomato fruit was the main aim. During the tomato-ripping process, the green tomato turns to red in several color stages (Ambrus et al., 2024). While the chlorophyll concentration decreases, the lycopene concentration increases. The sugar and the acid increase parallel to lycopene. The RGB camera can capture the process but needs to convert HSV color space to identify the tomato. The successful identification depends on the direct illumination volume. The experiment contains 4 ripe tomatoes and 15 different artificial illumination levels. The measurements show that the results are similar to or constantly above 3,000 lx illumination. However, under 3,000 lx, the detected size of tomatoes looks smaller and smaller depending on the weakness of illumination. Around 1,600 lx, it is possible to measure only half of the real size of the tomato. It shows that using the right amount of light is crucial to precise measurement in HSV color space. This research highlights the critical importance of proper illumination in ensuring accurate image analysis for tasks like industrial tomato segmentation. It emphasizes the need for adaptable lighting solutions, particularly in varying weather conditions, and the balance between adequate light and energy efficiency.
Open Access: Yes
DOI: 10.3303/CET24114139
Challenges of ecocentric sustainable development in agriculture with special regard to the internet of things (IoT), an ICT perspective
Publication Name: Progress in Agricultural Engineering Sciences
Publication Date: 2023-12-20
Volume: 19
Issue: 1
Page Range: 113-122
Description:
“Feed the global population and regenerate the planet.” The conditions necessary for the implementation of the above commonly used slogan did not exist 10–15 years ago. We did not have access to the information and databases that would have allowed us to increase yields for the purpose of feeding the growing population. While increasingly meeting sustainability requirements and regenerating the Earth. Anthropocentrism, the belief that humans are superior to everything else, benefits humans by exploiting human greed and ignorance, which is a dead end for both individuals and societies. Only humans can ignore the dynamic equilibrium processes of nature and disregard the consequences that adversely affect future generations. Ecocentric agricultural practices have several prerequisites. It is important for the academic sphere to recognize its significance. Another fundamental challenge is the continuous monitoring of the production unit and its close and distant environment for the purpose of decision preparation using Big Data. The Internet of Things (IoT) is a global infrastructure that represents the network of physical (sensors) and virtual (reality) “things” through interoperable communication protocols. This allows devices to connect and communicate using cloud computing and artificial intelligence, contributing to the integrated optimization of the production system and its environment, considering ecocentric perspectives. This brings us closer to the self-decision-making capability of artificial intelligence, the practice of machine-to-machine (M2M) interaction, where human involvement in decision-making is increasingly marginalized. The IoT enables the fusion of information provided by deployed wireless sensors, data-gathering mobile robots, drones, and satellites to explore complex ecological relationships in local and global dimensions. Its significance lies, for example, in the prediction of plant protection. The paper introduces small smart data logger robots, including the Unmanned Ground Vehicles (robots) developed by the research team. These can replace sensors deployed in the Wireless Sensor Net (WSN).
Open Access: Yes
Challenges of sustainable agricultural development with special regard to Internet of Things: Survey
Publication Name: Progress in Agricultural Engineering Sciences
Publication Date: 2022-12-02
Volume: 18
Issue: 1
Page Range: 95-114
Description:
If we want to increase the efficiency of precision technologies to create sustainable agriculture, we need to put developments and their application on a new footing; moreover, a general paradigm shift is needed. There is a need to rethink close-At-hand and far-off innovation concepts to further develop precision agriculture, from both an agricultural, landscape, and natural ecosystem sustainability perspective. With this, unnecessary or misdirected developments and innovation chains can be largely avoided. The efficiency of the agrotechnology and the accuracy of yield prediction can be ensured by continuously re-planning during the growing season according to changing conditions (e.g., meteorological) and growing dataset. The aim of the paper is to develop a comprehensive, thought-provoking picture of the potential application of new technologies that can be used in agriculture, primarily in precision technology-based arable field crop production, which emphasizes the importance of continuous analysis and optimisation between the production unit and its environment. It should also be noted that the new system contributes to reconciling agricultural productivity and environmental integrity. The study also presents research results that in many respects bring fundamental changes in technical and technological development in field production. The authors believe that treating the subsystems of agriculture, landscape, and natural ecosystem (ALNE) as an integrated unit will create a new academic interdisciplinarity. ICT, emphasizing WSN (Wireless Sensor Network), remote sensing, cloud computing, AI (Artificial Intelligence), economics, sociology, ethics, and the cooperation with young students in education can play a significant role in research. This study treats these disciplines according to sustainability criteria. The goal is to help management fulfil the most important expectation of reducing the vulnerability of the natural ecosystem. The authors believe that this article may be one of the starting points for a new interdisciplinarity, ALNE.
Open Access: Yes
Digital Twin of Drone-based Protection of Agricultural Areas
Publication Name: 2022 IEEE 1st International Conference on Internet of Digital Reality Iod 2022
Publication Date: 2022-01-01
Volume: Unknown
Issue: Unknown
Page Range: 99-104
Description:
Protecting agricultural fields, like crops, vineyards, and husbandry areas, has been a difficult challenge since historical times. Classical methods to prevent intrusion are often destructive to wild and domestic animals alike. Even more current nondestructive systems, like camera-based systems are attributed to specific problems related to environmental or technological issues. Furthermore, verifying the effectiveness of installed systems is difficult, as the triggering situations are unmanageable and typically occur unsupervised. This paper presents a complex vision-based intrusion detection system to overcome these problems and further proposes more extensive control and flexibility on the development process. The solution provides a workflow integrating Digital Reality methods into the system development by creating a digital twin of the drone and its surrounding environment in a general-purpose robotic simulator. With this simulation, the triggering events and environmental effects can be easily emulated, such as a wild animal entering the area of interest. The solution also focuses on incorporating new 5G info-communication networks on handling communication between the intrusion detection system and the base station in a distributed manner.
Open Access: Yes
