Improved Waste and Litter Image Classification with CBAM-integrated VGG-16 Vs VGG-16

The effective classification and segregation of waste materials are critical steps toward achieving sustainable waste management and environmental preservation. However, conventional manual sorting methods are time-consuming, labor-intensive, and prone to human error, resulting in inefficiencies and ecological hazards. To address these challenges, this study introduces a novel deep learning architecture that enhances the classical VGG-16 model by integrating the Convolutional Block Attention Module (CBAM) for multi-class classification of litter and waste images. The proposed framework leverages the publicly available Kaggle Garbage Classification dataset, which consists of 7,440 balanced images categorized into 12 waste classes: battery, biological, brown glass, cardboard, clothes, green glass, metal, paper, plastic, shoes, trash, and white glass. In this work, CBAM modules are strategically incorporated into the VGG-16 architecture at multiple network stages—specifically after the last and second-to-last dense layers—to investigate the influence of spatial and channel-wise attention mechanisms on model performance. This modification allows the network to adaptively focus on the most informative features, thereby improving discriminative learning and reducing irrelevant background noise. The CBAM-enhanced VGG-16 model is comprehensively evaluated and compared with the baseline VGG-16 model to assess their classification efficiency. Experimental results demonstrate that the integration of CBAM significantly improves classification accuracy, indicating superior capability in capturing fine-grained visual details and contextual information essential for differentiating between visually similar waste categories. The findings confirm the robustness and effectiveness of the proposed CBAM-VGG-16 framework for real-world waste management applications, suggesting its potential as a scalable and automated solution for intelligent waste segregation systems.