Garbage Detection on The Water Surface Based on Deep Learning

摘要

Based on an intelligent water surface cleaning underwater vehicle, this paper applies deep learning to the detection of water surface garbage in order to realize the automatic detection and classification of water surface garbage. The water surface garbage image is collected by the underwater robot through the camera for image preprocessing and in-depth learning based on neural network. The image preprocessing includes three stages: Canny operator edge extraction, image segmentation and histogram averaging image enhancement. The ANNs algorithm based on convolutional neural network is used for the object recognition, and the model is trained by transfer learning method. The experimental results show the recognition effect of image processing.