Apple orchard production estimation using deep learning strategies: A comparison of tracking-by-detection algorithms

Simple item page
dc.coverageDOI: 10.1016/j.compag.2022.107513
dc.creatorVillacrés, Juan
dc.creatorViscaino, Michelle
dc.creatorDelpiano, José
dc.creatorVougioukas, Stavros
dc.creatorAuat Cheein, Fernando
dc.date2023
dc.date.accessioned2025-11-18T19:48:58Z
dc.date.available2025-11-18T19:48:58Z
dc.description<p>The automated detection and counting of fruit in tree canopies is a key component of yield estimation systems, which are indispensable for the precision management of modern orchards. Detection and counting tasks in agricultural environments are not trivial because of challenges such as characteristics of the tree canopies, occlusion caused by leaves and the lighting conditions, among other factors. With the aim of identifying which algorithm is more suitable for yield estimation, we present a comprehensive comparison of tracking-by-detection algorithms, applied to apple counting. The tracking strategies evaluated were Kalman Filter, Kernelized Correlation Filter, Simple Online Real-Time Tracking, Multi Hypothesis Tracking, and Deep Simple Online Real-Time Tracking. The five tracking algorithms were further assessed on two novel databases constructed for this research in Multiple Object Tracking MOT format. After a sensitivity analysis of the trackers, the results show that the most robust approach is the Multiple Hypothesis Tracking, followed by the Deep Simple Online Realtime (DeepSORT), with a MOT accuracy of 97.00% and 93.00%, respectively, when having perfect detection. However, in an application case including a deep learning-based detection stage, the DeepSORT tracker obtains the lowest counting error, which on average for all videos is 20.07% and 31.52% when using YoloV5 and Faster R-CNN as detection strategies. Statistically similar results were obtained using the Kalman Filter with a counting error of 20.5% and 31.9% when detecting fruit with YoloV5 and Faster R-CNN.</p>eng
dc.descriptionThe automated detection and counting of fruit in tree canopies is a key component of yield estimation systems, which are indispensable for the precision management of modern orchards. Detection and counting tasks in agricultural environments are not trivial because of challenges such as characteristics of the tree canopies, occlusion caused by leaves and the lighting conditions, among other factors. With the aim of identifying which algorithm is more suitable for yield estimation, we present a comprehensive comparison of tracking-by-detection algorithms, applied to apple counting. The tracking strategies evaluated were Kalman Filter, Kernelized Correlation Filter, Simple Online Real-Time Tracking, Multi Hypothesis Tracking, and Deep Simple Online Real-Time Tracking. The five tracking algorithms were further assessed on two novel databases constructed for this research in Multiple Object Tracking MOT format. After a sensitivity analysis of the trackers, the results show that the most robust approach is the Multiple Hypothesis Tracking, followed by the Deep Simple Online Realtime (DeepSORT), with a MOT accuracy of 97.00% and 93.00%, respectively, when having perfect detection. However, in an application case including a deep learning-based detection stage, the DeepSORT tracker obtains the lowest counting error, which on average for all videos is 20.07% and 31.52% when using YoloV5 and Faster R-CNN as detection strategies. Statistically similar results were obtained using the Kalman Filter with a counting error of 20.5% and 31.9% when detecting fruit with YoloV5 and Faster R-CNN.spa
dc.identifierhttps://investigadores.uandes.cl/en/publications/8c0683fd-5826-4a49-8e95-9558fa165bef
dc.identifier.urihttps://repositorio.uandes.cl/handle/uandes/55834
dc.languageeng
dc.rightsinfo:eu-repo/semantics/restrictedAccess
dc.sourcevol.204 (2023)
dc.subjectApple counting
dc.subjectFaster R-CNN
dc.subjectMulti-object tracking
dc.subjectTracking-by-detection
dc.subjectYoloV5
dc.titleApple orchard production estimation using deep learning strategies: A comparison of tracking-by-detection algorithmseng
dc.typeArticleeng
dc.typeArtículospa
Files
Collections
PURE