Waste Classification using VGG-16 and ResNet-50
Sakshi Dharwadkar1, Vasundhara Patil2, Amulya Patil3, Soniya Chougule4,Prof. Plasin Dias5
1234Student, Department of Electronics and Communication Engineering, KLS Vishwanathrao Deshpande Institute of Technology, Haliyal, Karnataka, India
5Asst.Prof., Department of Electronics and Communication Engineering, KLS Vishwanathrao Deshpande Institute of Technology, Haliyal, Karnataka, India
Abstract - The increase in solid waste has made waste management a major challenge in today’s world. Manual waste segregation is slow and often inaccurate because it depends on human effort, which can lead to mistakes. To overcome these problems, automated and software-based waste classification systems are becoming increasingly important.
Convolutional Neural Networks (CNN) are used in this project to create a waste classification system. Two deep learning models, VGG-16 and ResNet-50, are used throughout the implementation in a Kaggle Notebook environment. The waste image dataset is gathered from Kaggle and preprocessed using data augmentation techniques, pixel value normalization, and image resizing. Next, training and testing sets are created from the dataset. The VGG-16 and ResNet-50 models are both trained to categorize waste images into various pre-established groups. Accuracy, loss graphs, precision values, and confusion matrices are used to assess the models' performance. The results show that ResNet-50 achieves better accuracy and lower loss because of its deeper network and residual learning capability, while VGG-16 offers a stable and dependable baseline performance. All things considered, this project shows an efficient and dependable software-based method for automated waste classification, which can promote improved recycling and waste disposal procedures.
Keywords: Waste Classification, Convolutional Neural Network (CNN), VGG-16, ResNet-50, Deep Learning, Kaggle Notebook, Image Processing, Solid Waste Management, Recycling Automation, Residual Learning, Transfer Learning, Supervised Learning, Computer Vision.
