Garbage Recognition Genius: Implementation of an Intelligent Waste Classification System and Interactive Platform Based on Enhanced ResNet50

Le résumé fourni par la source

In response to the actual demands of automated sorting of urban domestic waste, and in view of the shortcomings of the traditional ResNet50 model in extracting features in complex environments and the low accuracy of small target waste recognition in the garbage classification task, this paper proposes an enhanced ResNet50 garbage classification recognition model that incorporates a multi-scale feature fusion module and a channel attention mechanism. By optimizing the structure of the residual module and the training strategy, the classification accuracy and generalization ability of the model are improved. At the same time, a corresponding visual interaction platform based on the PyQt5 framework is developed to realize the full process functions of garbage image upload, model inference, and real-time display of classification results. Experimental results show that on the public garbage classification dataset, the Top-1 classification accuracy of the proposed enhanced ResNet50 model reaches 96.27%, which is 3.15 percentage points higher than that of the original ResNet50 model. The precision, recall rate, and F1 value have also been significantly improved. The ablation experiments fully verify the effectiveness of each enhancement module. The inference and result display response time for a single garbage image of the developed interaction platform is less than 200ms, demonstrating good practicality and user experience, and providing technical support and solutions for the intelligent garbage sorting in urban sanitation scenarios.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.