Flexible Dual-Branch Siamese Network: Learning Location Quality Estimation and Regression Distribution for Visual Tracking
IEEE Transactions on Computational Social Systems
1.result
Dataset | SiamFDB | |
OTB100 | Success | 70.4 |
Precision | 91.9 | |
UAV123 | Success | 64.9 |
Precision | 84.0 | |
LaSOT | Success | 52.7 |
Norm precision | 60.9 | |
Precision | 54.0 | |
GOT10k | AO | 63.5 |
SR0.5 | 73.8 | |
SR0.75 | 51.0 | |
VOT2019 | EAO | 31.7 |
Accuracy | 60.3 | |
Robustness | 42.6 |
2. Environment setup
This code has been tested on Ubuntu 18.04, Python 3.7, Pytorch 1.7.1, CUDA 11. Please install related libraries before running this code:
pip install -r requirements.txt
Download the pretrained model:
The model has been trained for testing and validation.
model code: xyqt
Download testing datasets and put them into test_dataset
directory. If you want to test the tracker on a new dataset, please refer to pysot-toolkit to set test_dataset.
Download pretrained backbones
Download pretrained backbones from google driver or BaiduYun (code: 7n7d) and put them into pretrained_models
directory.
3.train
To train the SiamFDB model, run train.py with the desired configs:
For OTB and VOT Benchmark.
cd /path/to/SiamFDB
export PYTHONPATH=./:$PYTHONPATH
python tools/train.py --cfg ./experiments/SiamFDB_r50/configOTBVOT.yaml
For UAV Benchmark
cd /path/to/SiamFDB
export PYTHONPATH=./:$PYTHONPATH
python tools/train.py --cfg ./experiments/SiamFDB_r50/config.yaml
For GOT10k Benchmark
cd /path/to/SiamFDB
export PYTHONPATH=./:$PYTHONPATH
python tools/train.py --cfg ./experiments/SiamFDB_r50_got10k/config.yaml
For LaSOT Benchmark
cd /path/to/SiamFDB
export PYTHONPATH=./:$PYTHONPATH
python tools/train.py --cfg ./experiments/SiamFDB_r50_lasot/config.yaml
4.test
cd /path/to/SiamFDB
export PYTHONPATH=./:$PYTHONPATH
python tools/test.py --config ./experiments/SiamFDB_r50/config.yaml --dataset UAV123 --snapshot ./models/UAV123.pth
5.eval
please refer to pysot-toolkit
6.tune
cd /path/to/SiamFDB
export PYTHONPATH=./:$PYTHONPATH
python tools/tune.py \
--dataset_root /path/to/dataset/root \ # dataset path
--dataset UAV123 \ # dataset name(OTB100, GOT10k, LaSOT, UAV123, VOT2016, VOT2018, VOT2019)
--snapshot ./models/UAV123.pth \ # tracker_name
--config ./experiments/SiamFDB_r50/config.yaml \ # config file
7.Cite
If you use SiamFDB in your work please cite our paper:
@ARTICLE{10034430,
author={Hu, Shuo and Zhou, Sien and Lu, Jinbo and Yu, Hui},
journal={IEEE Transactions on Computational Social Systems},
title={Flexible Dual-Branch Siamese Network: Learning Location Quality Estimation and Regression Distribution for Visual Tracking},
year={2023},
volume={},
number={},
pages={1-9},
doi={10.1109/TCSS.2023.3235649}
}