ultrayolo
Tensorflow 2 implementation of yolo version 3
- Free software: Apache Software License 2.0
- Documentation: https://ultrayolo.readthedocs.io.
Features
ultrayolo implements YOLO object detection paper with the following backbones:
- Darknet: the original architecture
- ResNet: ['ResNet50V2', 'ResNet101V2', 'ResNet152V2']
- DenseNet: ['DenseNet121', 'DenseNet169', 'DenseNet201']
- MobileNetV2: ['MobileNetV2']
Examples
The following notebooks show how to use the library:
- Predict using the official Yolo checkpoint
- Predict using a custom trained model
- Explaination about how the Yolo loss works
- How to train the model
- How to modify the data augmentation pipeline
TODO
- [ ] add support to tf.data to load datasets
- [ ] add mean average precision evaluation script
Credits
This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.