This repository contains the code for the work on Optical Flow Estimation using Hybrid Neural Network. The code below is derived from the original implementation by Parts of this code were derived from chan8972/Spike-FlowNet.
Clone the repository using:
git clone https://github.com/siddux/spikeOpticalFlow.git
Create a conda environment using the environment.yml file:
conda env create -f environment.yml
Activate the conda environment:
conda activate spikeopticalflow
Download datasets folder and include in this directory
The basic syntax for training the neural network is:
python3 main.py
The pretrained model can be found in /pretrain folder.
The basic syntax for testing is:
python3 main.py --evaluate --pretrained='./pretrain/checkpoint_dt1.pth.tar' --render --save-images
Another commands are available to change running options in the main file.