Semantic segmentation model for segmenting sidewalks from other object in an image.<br> Utilizes U-Net with Resnet34 backbone for transfer learning.<br> Trained on 512x512 images and expects images with even dimensions.<br>