Semantic-Segmentation-T3P2. Semantic segmentation uses transposed convolutions in the decoder part to segment the images in classes. This projects implements a FCN8 architecture to detect the roads in the image.