In this project, I took images from the Oxford IIIT Pet dataset , built a CNN model for semantic segmentation, trained it to predict the class label for each pixel in the input image. The Oxford IIIT Pet dataset is a 37 category pet image dataset with roughly 200 images for each class which have been split randomly into 50 for training, 50 for validation, and 100 for testing . The ground truth data has annotations for species, head position, and pixel segmentation for each image.