Transfer Learning with Inception V3 for Bottle Type Classification in Recycling: A Deep Learning Framework

The automated classification of waste and hence dumping has become an essential responsibility in the pursuit of efficient waste management. This work focuses on applying Transfer Learning with the InceptionV3 convolutional neural network architecture for the binary categorization of bottle-type rubbish. For a variety of picture classification tasks, InceptionV3 offers a reliable feature extractor by utilising pre-trained weights from ImageNet. The top layers of InceptionV3 are altered in this work to better fit the particular purpose of classifying rubbish that looks like bottles. The dataset that includes pictures of bottles rubbish items is used to train the improved model. By utilising the information obtained from the pre-trained model, transfer learning enables quicker convergence and better results on the classification job. To test the trained model’s efficacy and accuracy in categorizing rubbish that resembles bottles, a different test dataset is used. The results show that Transfer Learning with InceptionV3 is effective at classifying binary rubbish, offering a potential method for automated waste sorting systems that aim to improve recycling and environmental sustainability.