Real-Time Classification and Detection of Garbage Based on Improved Yolov5 and Embedded System

Garbage classification is an important environmental issue, and accurate sorting of different types of waste is critical for effective recycling and waste management. Deep learning technology has great potential for garbage classification, and with the growing importance of waste management, it is highly likely that the role of image recognition in this field will continue to increase in the future. However, most of the existing garbage classification and detection methods only consider the detection accuracy of garbage, which are not real-time and can hardly meet the practical application of garbage classification. To address this problem, a new real-time classification and detection of garbage based on improved yolov5 and adapted to an embedded system from UAV images is proposed, which is named as G-YOLOv5s. First, residual dilated convolution module(Res-DConv) is employed to extract the spatial features, which can increase the receptive field. Then, a feature fusion module(BDSCAM) is designed to enhance the expressive ability of object feature, which could improve the classification performance of detector. The localization task is performed using a Double-Head method, which is an integration system of fully connected and convolutional heads for bounding box regression and classification. The proposed G-YOLOv5s method was evaluated using Huawei Garbage Classification Challenge Cup dataset. We found that the experimental results demonstrate that the proposed method is high accuracy and efficient for garbage classification and detection on the embedded system. This approach is robust and suitable for practical application in garbage classification scenarios.