This repository provides PyTorch implementations of popular semantic segmentation models such as FCN, U-Net, PSPNet, ICNet, FRRN, Link-Net, and SegNet. It includes support for loading pretrained weights and ships with data loaders for standard segmentation benchmarks including CamVid, Pascal VOC, ADE20K, Cityscapes, and NYUDv2. The implementations focus on fully convolutional architectures for per-pixel classification in images.