Abstract: Garbage detection can reduce resource waste and alleviate environmental pollution, which is of great significance to environmental protection. This paper proposes a multi-scale garbage detection method with brightness enhancement to solve the problem of false recognition and missing detection in dim images and small objects. First, in the brightness enhancement mode, the skip connection is used to enhance the correlation of garbage image features at different levels, which solves the problem of garbage misrecognition under dim conditions. Then, in the multi-scale garbage detection module, dense connections are used to integrate the features of different scales, which improves the learning ability of fine garbage features and solves the problem of missing detection of fine garbage. The proposed method achieved 96.62% and 93.81% of the maps on self-made and public datasets. Experimental results show that this method can solve the problem of false recognition and missing detection in dim image and small object, and is superior to existing mainstream methods such as YOLOv4.