Detecting Biodegradable and Non-biodegradable Image Using CNN and Deep Learning

To distinguish between biodegradable and non-biodegradables, different methodologies and models are implemented in this study using image processing. We experimented with Convolutional Neural Networks (CNN), U-Net, and Support Vector Machines (SVM). CNN outperformed the other models obtaining a training accuracy of 0.9397, which is approximately 94%, and a validation accuracy of 0.9165, which can be rounded off to 92%. SVM on the other hand obtained an accuracy of 92%. We assessed our model's accuracy and our approach was able to predict well whether the object is biodegradable or not based on the input of a single image. To make sure that it can recognize the primary objective from a single image, we went through our models extensively. The identification, tracking, sorting, and processing of biodegradable and non-biodegradable items might be aided by our model when it will be improved.