Following tasks are done:

1. Download the MS-COCO annotations and create a histogram of object classes.
2. Next, extract two sets of 100 images in which the a) most often occurring object class is present, and b) least often occurring object class is present.
3. For each of those images, replace the background (where no object is present) with a black-colored background.
4. Find a tutorial to train a small GAN using PyTorch on the original sets of images and “ blacked-out” image sets.
5. Report your findings.

I have referenced below mentioned online materials for completing the task (also mentioned in notebook):

1. https://www.analyticsvidhya.com/blog/2019/07/computer-vision-implementing-mask-r-cnn-image-segmentation/
2. https://towardsdatascience.com/master-the-coco-dataset-for-semantic-image-segmentation-part-1-of-2-732712631047
3. https://towardsdatascience.com/master-the-coco-dataset-for-semantic-image-segmentation-part-2-of-2-c0d1f593096a
4. https://pytorch.org/tutorials/beginner/dcgan_faces_tutorial.html