https://github.com/Zaabon/spiking_yolo

https://github.com/Zaabon/spiking_yolo

Last synced: 12 days ago
JSON representation

• Host: GitHub
• URL: https://github.com/Zaabon/spiking_yolo
• Owner: Zaabon
• License: gpl-3.0
• Created: 2021-02-08T11:13:14.000Z (over 3 years ago)
• Default Branch: master
• Last Pushed: 2021-03-24T11:43:41.000Z (over 3 years ago)
• Last Synced: 2024-02-29T09:35:12.353Z (4 months ago)
• Language: Python
• Size: 44.3 MB
• Stars: 11
• Watchers: 2
• Forks: 0
• Open Issues: 2
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Lists

• awesome-yolo-object-detection - Zaabon/spiking_yolo

README

        
This project is a combined neural network utilizing an spiking CNN with backpropagation and YOLOv3 for object detection. 

## Requirements

Python 3.8 or later with all [requirements.txt](https://github.com/Zaabon/spiking_yolo/blob/95477ded20c2dc8c69115b7de9063ec1da13204d/requirements.txt) dependencies installed, including `torch>=1.6`. In addition you will need to have setup Jupyter with CUDA support for GPU. 

## Running the project

Download the dataset N-Caltech101 from https://www.garrickorchard.com/datasets/n-caltech101 and unzip in the project root. 

With Jupyter, run [data_processing.ipynb](https://github.com/Zaabon/spiking_yolo/blob/95477ded20c2dc8c69115b7de9063ec1da13204d/data_processing.ipynb). This might take a while, you might not need to process all data. 

The full network can be run through [spiking_yolov3.ipynb](https://github.com/Zaabon/spiking_yolo/blob/95477ded20c2dc8c69115b7de9063ec1da13204d/spiking_yolov3.ipynb) while only the original YOLOv3 is runnable through [original_yolov3.ipynb](https://github.com/Zaabon/spiking_yolo/blob/95477ded20c2dc8c69115b7de9063ec1da13204d/original_yolov3.ipynb). 

Code for spiking solution is modified from https://github.com/yjwu17/BP-for-SpikingNN and can be found in the directory SpikingNN. 

Most of main YOLOv3 code can be found in [train.py](https://github.com/Zaabon/spiking_yolo/blob/95477ded20c2dc8c69115b7de9063ec1da13204d/train.py) which is originally from https://github.com/ultralytics/yolov3