UNet built from scratch for semantic segmentation.
The UNet is built exactly as suggested in the original paper: https://arxiv.org/abs/1505.04597.
It is built using the powerful PyTorch library, which is object-oriented and utilizes pythonic code!
The VOC dataset is used to test the model, but any other dataset can be used with the requirement of a class in the folder dataset.
- torch
- torchvision
- numpy
- pillow
- scikit-learn
- tqdm