Abstract
In order to quickly identify and locate passion fruits with different maturity in natural environment, a multi-class passion fruit fast detection algorithm based on YOLOv3 is proposed. Firstly, the data is preprocessed and made into VOC2007 data set format. Secondly, the Densenet network is added to YOLOv3 feature extraction module for enhancing the feature propagation of convolutional layers. Finally, the multi-scale prediction is reduced to single-scale prediction in YOLOv3, and medium- sized object detectors are retained among large, medium, and small object detectors. The experimental results show that the method can effectively detect passion fruit with different maturity in natural environment.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call