Object Detection and Segmentation

A Machine Learning project to detect and segment a set of MNIST RGB images (images of digits from 0 to 9). It determines the boundary boxes of digits in an image and segments them from the background.

Overall accuracy: 99.6%

Link to download dataset and saved model: https://drive.google.com/drive/folders/1r4QomZ9X9FSWZH9OUeiFDqjBzuSOpjzZ?usp=sharing

To run the program:

1. Download the dataset and the saved model. Place them in the same directory with the python files.

2. Run main.py.