Banana Pseudostem Visual Detection Method Based on Improved YOLOV7 Detection Algorithm

Detecting banana pseudostems is an indispensable part of the intelligent management cultivation, which can be used in settings such as counting and smart fertilization. In complex environments, dense occlusion pose a significant challenge for detection. This paper proposes improved YOLOV7 deep learning object detection algorithm, YOLOV7-FM, detecting with different growth conditions. loss optimization model, Focal introduced, to optimize problematic training that are sheltered, so improve recognition rate challenging samples. data augmentation Mixup used, model’s generalization ability similar features environments. compares AP (average precision) inference speed YOLOV7-FM algorithm YOLOX, YOLOV5, YOLOV3, Faster R-CNN algorithms. The results show higher than those models compared, average time 8.0 ms per image containing 81.45%. model achieve fast accurate pseudostems.