Pytorch-UNet and tf_unet
About Pytorch-UNet
milesial/Pytorch-UNet
PyTorch implementation of the U-Net for image semantic segmentation with high quality images
Supports mixed precision training (FP16) and automatic gradient scaling for memory efficiency on modern GPUs, plus multiclass segmentation tasks beyond the original Carvana dataset. Includes Weights & Biases integration for real-time training visualization and a pretrained model loadable via torch.hub, with Docker containerization for reproducible environments.
About tf_unet
jakeret/tf_unet
Generic U-Net Tensorflow implementation for image segmentation
Implements the full U-Net architecture with skip connections for pixel-wise dense prediction across arbitrary imaging domains. Built on TensorFlow 1.x with demonstrated applications in radio astronomy (RFI mitigation), medical imaging, and synthetic pattern detection. Provides modular, domain-agnostic design allowing direct application to custom datasets without architecture modifications.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work