Darkeras: Executing YOLOv3/YOLOv4 Object Detection on Keras with Darknet Pre-trained Weights

This project is to execute YOLOv3/YOLOv4 object detector on Keras using Darknet pre-trained weights.

Dependencies

• OpenCV
• Tensorflow 2.3.0
• Keras 2.4.0
• Easydict
• Matplotlib

You can simply run:

$ pip install -r requirements.txt

Supported Models

• YOLOv3
• YOLOv4

How to Use

1. Run with Notebook

• Put pre-trained weights from official Darknet website or your trained weights into "weights" folder (If you use your model trained on your customed dataset, please change NUM_CLASS and ANCHORS in the notebooks)
• Run YOLOv3: darkeras-yolov3.ipynb
• Run YOLOv4: darkeras-yolov4.ipynb

2. Run with Python Scripts

• Convert Darknet (format: weight_file.weights) to Keras (format: weight_file.h5): to be updated
• Run YOLOv3: to be updated

$ python detect_yolov3.py

• Run YOLOv4: to be updated

$ python detect_yolov4.py

Detection Results

• Input

• YOLOv3

• YOLOv4

Citation

Please cite this repo if it's helpful for your study.

@misc{tran2021darkeras,
  author = {Tran, Le-Anh},
  title = {Darkeras: Executing YOLOv3/YOLOv4 Object Detection on Keras with Darknet Pre-trained Weights},
  year = {2021},
  publisher = {GitHub},
  journal = {GitHub repository}
}

Have fun!

LA Tran

August 2021