Focus-RCNet: a lightweight recyclable waste classification algorithm based on focus and knowledge distillation

• Published in: Visual computing for industry, biomedicine and art Vol. 6; no. 1; pp. 19 - 9
• Main Authors: Zheng, Dashun, Wang, Rongsheng, Duan, Yaofei, Pang, Patrick Cheong-Iao, Tan, Tao
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 11.10.2023; Springer Nature B.V; SpringerOpen
• Subjects: Accuracy; Algorithms; Artificial intelligence; Artificial neural networks; Attention; Automation; CAE) and Design; Classification; Computer Graphics; Computer Graphics International 2023; Computer Imaging; Computer Science; Computer-Aided Engineering (CAD; Continuous improvement; Deep learning; Distillation; Efficiency; Electronic waste; Environmental impact; Household wastes; Image classification; Image Processing and Computer Vision; Knowledge; Knowledge distillation; Lightweight; Mathematical models; Media Design; Neural networks; Original; Original Article; Parameters; Pattern Recognition and Graphics; Recycling; Standard of living; Support vector machines; Vision; Waste classification; Waste recycling; Waste classification; Knowledge distillation; Lightweight; Waste recycling; Attention
• ISSN: 2524-4442; 2096-496X; 2524-4442
• DOI: 10.1186/s42492-023-00146-3

Waste pollution is a significant environmental problem worldwide. With the continuous improvement in the living standards of the population and increasing richness of the consumption structure, the amount of domestic waste generated has increased dramatically, and there is an urgent need for further treatment. The rapid development of artificial intelligence has provided an effective solution for automated waste classification. However, the high computational power and complexity of algorithms make convolutional neural networks unsuitable for real-time embedded applications. In this paper, we propose a lightweight network architecture called Focus-RCNet, designed with reference to the sandglass structure of MobileNetV2, which uses deeply separable convolution to extract features from images. The Focus module is introduced to the field of recyclable waste image classification to reduce the dimensionality of features while retaining relevant information. To make the model focus more on waste image features while keeping the number of parameters small, we introduce the SimAM attention mechanism. In addition, knowledge distillation was used to further compress the number of parameters in the model. By training and testing on the TrashNet dataset, the Focus-RCNet model not only achieved an accuracy of 92 % but also showed high deployment mobility.