A simple experiment to test and evaluate the Segmentation algorithm to detect and segment the firearms.
All the code and python notebook which contains my project named "Firearms Segmentation using Mask R-CNN and Transfer Learning". This project was done as a term project for my Master's ML Class. A custom Mask R-CNN was trained on IMFDB Dataset to detect and segment the melanoma in skin dermoscopic images. I used ResNet 50 for transfer learning and http://www.robots.ox.ac.uk/~vgg/software/via/ was used as annotation tool. Code was used by @waleedabdullah. Which I modified as per my requirements.
If you want to develop a classifier for your own custom dataset please see links below