This repository provides a Semantic Segmentation Suite built in TensorFlow for implementing, training, and testing semantic segmentation models. It includes multiple encoder-decoder architectures (SegNet, Mobile UNet, RefineNet) with various backbone networks (MobileNetV2, ResNet, InceptionV4). The suite supports data augmentation, training and testing modes, and evaluation metrics including precision, recall, F1 score, per-class accuracy, and mean IoU.