Detecting Apples in Orchards Using YOLOv3 and YOLOv5 in General and Close-Up Images.
2020
A machine vision system for apple harvesting robot was developed based on the YOLOv3 and the YOLOv5 algorithms with special pre- and post-processing and the YOLOv3 equipped with special pre- and post-processing procedures is able to achieve an a share of undetected apples (FNR) at 9.2% in the whole set of images, 6,7% in general images, and 16,3% in close-up images. A share of objects mistaken for apples (FPR) was at 7.8%. The YOLOv5 can detect apples quite precisely without any additional techniques, showing FNR at 2.8% and FPR at 3.5%.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
32
References
6
Citations
NaN
KQI