Performance Comparison of YOLO Setups for Agriculture Machine Surrounding Monitoring
Fragment książki (Rozdział monografii pokonferencyjnej)
MNiSW
20
Poziom I
| Status: | |
| Autorzy: | Tarasiuk Krzysztof, Mystkowski Arkadiusz, Ostaszewski Michał, Majka Andrzej, Czarnigowski Jacek |
| Dyscypliny: | |
| Aby zobaczyć szczegóły należy się zalogować. | |
| Wersja dokumentu: | Drukowana | Elektroniczna |
| Język: | angielski |
| Strony: | 1 - 6 |
| Scopus® Cytowania: | 0 |
| Bazy: | Scopus | IEEE Xplore |
| Efekt badań statutowych | NIE |
| Materiał konferencyjny: | TAK |
| Nazwa konferencji: | 26th International Carpathian Control Conference |
| Skrócona nazwa konferencji: | 26th ICCC 2025 |
| URL serii konferencji: | LINK |
| Termin konferencji: | 19 maja 2025 do 21 maja 2025 |
| Miasto konferencji: | Starý Smokovec |
| Państwo konferencji: | SŁOWACJA |
| Publikacja OA: | NIE |
| Abstrakty: | angielski |
| Object detection combines the localization of objects in an image with their classification, which is challenging due to the complexity of the environment, the size and variety of objects, and the limited computational power. This paper presents an implementation of object recognition for agricultural machinery, which is a key element in developing autonomous control and safety systems. The research includes two selected and trained models of algorithms for recognizing objects in images: two versions of the You Only Look Once (YOLO) algorithm, which are characterized by fast operation, but have problems in recognizing small objects and with different landscapes. The trained models were implemented on a low-cost processor system. The simulation tests are carried out with the entire system installed in an agricultural machine. The results show that the processor system makes predictions based on the camera image and provides information about accurately recognized objects. In addition, information about the position of a particular object in the image is given in pixels. The results show mean average precision for YOLO v8s equals 0.879. This network also detected harvesters with 97% accuracy and tractors with an accuracy of 94%. |
