This project aims to mitigate the classification of trash objects by developing a model that automatically detects different types of waste products into predefined classes. The solution begins with implementing Faster R-CNN model to estimate the boundary of each category of object present within an image using a confidence score (IOU Metric). This further extends to implement Mask R-CNN for image segmentation to detect the type of waste.
- According to the studies published by the "Recycling rate of municipal solid waste in the United States 1960-2018. Published by Ian Tiseo, Mar 30, 2022 ", only 30% of recyclable materials actually get recycled.
- Automated system needed to improve efficiency and create a sustainable process to manage waste.
- Instance segmentation is challenging as it requires the correct detection of all objects in an image while also precisely segmenting each instance.
Kaiming He, Georgia Gkioxari, Piotr Dollar, Ross Girshick, ‘Mask R-CNN’
- The data was scraped from Google Images using the Simple Image Downloader library.
- The images are broadly classified into 6 classes such as Metal, Glass, Paper, Organic, E-Waste and Medical.
- We have: 856 examples, 685 are training and 171 testing. (80-20 split).
