High Accuracy Lightweight Image Classification Using An Improved Yolo And Vgg16

Abstract

Waste Classification Plays A Crucial Role In Sustainable Waste Management By Categorizing Materials Based On Their Type To Ensure Proper Disposal And Recycling. Traditional Waste Sorting Methods, Which Rely Heavily On Manual Labor, Are Time-consuming, Error-prone, And Inefficient At Scale. The Exponential Increase In Global Waste Production Necessitates More Accurate And Automated Solutions. This Paper Proposes A Smart Waste Classification System That Integrates Two Deep Learning Paradigms: The VGG16 Convolutional Neural Network (CNN) For High-accuracy Feature-based Classification And The YOLO (You Only Look Once) Framework For Real-time Object Detection. Waste Images Are Preprocessed Through Normalization, Noise Filtering, And Data Augmentation Before Being Fed Into The Dual-model Pipeline. The VGG16 Model, Leveraging Transfer Learning From ImageNet Weights, Classifies Waste Into Six Categories Cardboard, Glass, Metal, Paper, Plastic, And Trash With High Precision. Concurrently, YOLO Identifies And Localizes Waste Items In Live Camera Feeds, Enabling Real-time Sorting Decisions Communicated Via Email And SMS Alerts. Experimental Results Demonstrate That The Integrated System Achieves Superior Classification Accuracy And Processing Speed Compared To Existing Single-model Approaches, Making It A Viable Solution For Automated Waste Management In Smart Cities, Recycling Facilities, And Industrial Environments.

Keywords