YOLO evaluation on custom dataset

main.ipynb DATASET

This is a yolo and yolo tiny evaluation (calculating precision-recall graph and mAP) on a simple road cctv footage dataset.

See this for more details. https://github.com/rahatzamancse/yolo-evaluation-on-custom-dataset/blob/master/main.ipynb

Motivation

I have made this repository for evaluating my this projects(traffic) model against yolo and yolo-tiny.

Requirements

• opencv
• tqdm
• pandas
• seaborn

Getting started

1. Download the dataset from kaggle https://www.kaggle.com/insaneshadowzaman/highway-cctv-footage-images#data.zip
2. Just go through the main.ipynb

-------- OR --------

Use the scripts manually :

python yolo.py --images input_images --yolo yolo-coco # or --yolo yolo-tiny-coco
python convert_to_yolo_format.py
python labelbox_json_reformatter.py
python calculate_mean_ap.py

Initial Configurations

Please set the model, CONFIDENCE and THRESHOLD variables in the In[2] of the main.ipynb file. All those variables are self explanatory if you are familiar with yolo models.

Contributing

The main reason to publish something open source, is that anyone can just jump in and start contributing to my project. So If you'd like to contribute, please fork the repository and use a feature branch. Pull requests are warmly welcome.

Links and References

Even though this information can be found inside the project on machine-readable format like in a .json file, it's good to include a summary of most useful links to humans using your project. You can include links like:

• Project homepage: https://github.com/rahatzamancse/yolo-evaluation-on-custom-dataset
• Repository: https://github.com/rahatzamancse/yolo-evaluation-on-custom-dataset.git
• Issue tracker: https://github.com/rahatzamancse/yolo-evaluation-on-custom-dataset
  • In case of sensitive bugs like security vulnerabilities, please contact rahatzamancse@gmail.com directly instead of using issue tracker. I value your effort.

Author

Rahat Zaman rahatzamancse@gmail.com Student at Department of Computer Science and Engineering Khulna University of Engineering & Technology, Khulna Bangladesh

Licensing

The code in this project is licensed under GNU GPLv3 license.