CB-retinanet

Class balanced retinanet with Visdrone dataset. Followed by Keras-retinanet, we change the focal loss to class-balanced focal loss(https://openaccess.thecvf.com/content_CVPR_2019/papers/Cui_Class-Balanced_Loss_Based_on_Effective_Number_of_Samples_CVPR_2019_paper.pdf) under retinanet and add VisDrone-2019 dataset as one of the trainingsets.

Installation

1. Clone this repository.
2. In the repository, execute pip install . --user. Note that due to inconsistencies with how tensorflow should be installed, this package does not define a dependency on tensorflow as it will try to install that (which at least on Arch Linux results in an incorrect installation). Please make sure tensorflow is installed as per your systems requirements.
3. Alternatively, you can run the code directly from the cloned repository, however you need to run python setup.py build_ext --inplace to compile Cython code first.
4. Optionally, install pycocotools if you want to train / test on the MS COCO dataset by running pip install --user git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI.

Training on VisDrone-019 dataset

# Running directly from the repository:
keras_retinanet/bin/train.py visdataset /path/to/VisDrone/dataset

More information please reference to keras-retinanet

Acknowledgement

This code is extended from the following repositories.

• keras-retinanet Thank the authors for releasing their codes. Please also consider citing their works.