This is the official implementation with training code for 'GradNet: Gradient-Guided Network for Visual Object Tracking' (ICCV2019 Oral). For more details, please refer to:
We propose a GradNet to update the template in single object tracking based on template information and gradients.
Results on OTB100
- Tensorflow
- CUDA 8.0 and cuDNN 6.0
- Python 2.7 / Python 3.6
- Data preparation: Please refer to https://github.com/bertinetto/siamese-fc for details and change the data paths in parameters.py.
- Please run
$(ROOT_PATH)/train.py
to get your own model.
Please run $(ROOT_PATH)/track.py
for demo.
Licensed under an MIT license.
If you find GradNet useful in your research, please kindly cite our paper:
@InProceedings{GradNet_ICCV2019,
author = {Peixia Li, Boyu Chen, Wanli Ouyang, Dong Wang, Xiaoyun Yang, Huchuan Lu},
title = {GradNet: Gradient-Guided Network for Visual Object Tracking},
booktitle = {ICCV},
month = {October},
year = {2019}
}
If you have any questions, please feel free to contact pxli@mail.dlut.edu.cn
Many parts of this code are adopted from other related works (tensorflow-siamese-fc)