Welcome to YOLOv9Face, a state-of-the-art face detection system based on the YOLOv9 architecture, specially trained on the WIDER FACE dataset to provide high accuracy and real-time performance across diverse scenarios.
- State-of-the-Art Accuracy: Utilizes the powerful YOLOv9 architecture, fine-tuned for face detection to ensure high precision across various environments.
- Real-Time Detection: Designed for efficiency, enabling real-time face detection on both high-end GPUs and edge devices.
- Robust and Reliable: Excellently handles challenging conditions such as low light, occlusions, and varied face orientations.
- Ease of Use: Simple setup and integration process, perfect for both standalone applications and integration into existing systems.
- Open Source and Customizable: Fully open-source for easy customization and further improvement on different datasets.
Ensure you have Python 3.8+ and pip installed on your system.
Clone this repository and install the required dependencies:
- git clone https://github.com/raoofaltaher/YOLOv9Face.git
- git clone https://github.com/WongKinYiu/yolov9
- cd YOLOv9Face
- pip install -r requirements.txt
The WIDER dataset comprises of more than 30k images with more than 390k faces, each with bouding box and other various label formats.
Dataset structure
${ROOT}
└── yolov9
└── datasets/
└── widerface/
└── train/
└── val/
└── original-widerface/
└── train/
└── images/
└── label.txt
└── val/
└── images/
└── label.txt
└── train2yolo.py
└── val2yolo.py
└── widerface.yaml
To prepare the data:
- Download the WIDER-FACE datasets.
- Download the annotation files from google drive.
Run the following commands:
python train2yolo.py datasets/original-widerface/train datasets/widerface/train
python val2yolo.py datasets/original-widerface datasets/widerface/valThese scripts will convert your annotation files to YOLO format, creating one .txt file per image. Each row in the file will represent a single object in the format: class x_center y_center width height.
To detect faces in an image using a pretrained model
- python detect.py --source path/to/your/image.jpg
@article{wang2024yolov9,
title={{YOLOv9}: Learning What You Want to Learn Using Programmable Gradient Information},
author={Wang, Chien-Yao and Liao, Hong-Yuan Mark},
booktitle={arXiv preprint arXiv:2402.13616},
year={2024}
}
@article{chang2023yolor,
title={{YOLOR}-Based Multi-Task Learning},
author={Chang, Hung-Shuo and Wang, Chien-Yao and Wang, Richard Robert and Chou, Gene and Liao, Hong-Yuan Mark},
journal={arXiv preprint arXiv:2309.16921},
year={2023}
}
- YOLOv9 - YOLOv9: Learning What You Want to Learn Using Programmable Gradient Information
- WIDER FACE - WIDER FACE: A Face Detection Benchmark
- YOLO5Face