Currently,research in the field of smart pepper harvesting remains unexplored,and one of the key technologies involves accurately identifying the maturity of pepper fruit spikes.This study employed two technical routes to determine pepper maturity.The first method was to establish a deep learning model for pepper target detection,followed by assessing pepper maturity based on the color characteristics of the pepper fruit spikes.The second method was to directly establish a deep learning model for determining pepper maturity.Both methods utilized five algorithms for comparison,including SSD,Faster R-CNN,YOLOv5s,YOLOv5m and YOLOv8m,to discriminate pepper fruit spikes maturity.The research results found that based on the YOLOv8m model,the first method achieved a maturity discrimination precision of 94.81％,while the second method demonstrated an excellent performance,with precision,recall rate and other metrics exceeding 98％,providing an important basis for the development of intelligent pepper harvesting robots.
维普
万方数据