Artificial Intelligence Based Approach to Identify Plastic Bottle for Environment Sustainability

Several classification problems have been resolved with computer vision and deep learning technology. Despite this, no dedicated work has been carried out to classify plastic bottles from non-bottle objects. Therefore, this paper is concerned to resolve issue to classify waste plastic bottle from non-bottle objects. Here a dataset of 52000 images of bottle and non-bottle has been used to classify different objects based on their size. Each image contains a single bottle and non-bottle object. In this paper, various pre-trained models have been employed to classify these objects based on their size. The pretrained models have been trained with bottle dataset and attained 97%, 92% and 91 % accuracy. The comparison shows that the ResNet-18 outperforms.