Pytorch CASENet implementation
This repository includes an untested PyTorch implementation of CASENet: Deep Category-Aware Semantic Edge Detection (https://arxiv.org/abs/1705.09759).
Usage:
Run train.py with argument --data-root pointing to the SBD root directory
To execute this, you must have Python 3.6.*, PyTorch, scikit-image, Numpy and Matplotlib installed. To accomplish this, we recommend installing the Anaconda Python distribution and use conda to install the dependencies, as it follows:
conda install matplotlib numpy scikit-image
conda install pytorch torchvision cuda80 -c soumithTo-do
- script for downloading and preparing dataset
- train and test on SBD
- train and test on Cityscapes