# Siamese-RPN-pytorch

**Repository Path**: kloov/Siamese-RPN-pytorch

## Basic Information

- **Project Name**: Siamese-RPN-pytorch
- **Description**: This is a re-implementation of Siamese-RPN with pytorch, which is CVPR2018 spotlight.
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2020-02-20
- **Last Updated**: 2020-12-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Siamese-RPN-pytorch
## Introduction
- [**Tensorflow Version**](https://github.com/makalo/Siamese-RPN-tensorflow.git) has been available by my classmates  [**makalo**](https://github.com/makalo). **If you have any question, please feel free to contact us.**    
- This is a re-implementation for [**High Performance Visual Tracking with Siamese Region Proposal Network**](http://openaccess.thecvf.com/content_cvpr_2018/papers/Li_High_Performance_Visual_CVPR_2018_paper.pdf) with PyTorch, which is accepted at CVPR2018.  
- Code_v1.0 is available for traning, you should change your dataset as **VOT** format(top-left point and w,h). 
  If there is a break in a sequence, ues "0,0,0,0" to replace the info of this frame.
- Dataset Tree 
```
-root/class1/img1.jpg
            /...
            /imgN.jpg
            /groundtruth.txt
```

## Citation
```
Paper: @InProceedings{Li_2018_CVPR,
author = {Li, Bo and Yan, Junjie and Wu, Wei and Zhu, Zheng and Hu, Xiaolin},
title = {High Performance Visual Tracking With Siamese Region Proposal Network},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2018}
}
```

## Getting Started
### Performance
<div align=center><img width="950" height="700" src="https://github.com/songdejia/siamese-RPN/blob/master/screenshot/test2.gif"/></div>

### Network introduction  
<div align=center><img width="730" height="360" src="https://github.com/songdejia/siamese-RPN/blob/master/screenshot/network.png"/></div>

### Environment  
- python=3.6  
- pytorch=0.4.0  
- cuda=9.0  
- shapely=1.6.4

### Download VOT2013 Dataset
```
wget http://data.votchallenge.net/vot2013/vot2013.zip 
```

### Download YouTube-BB Data
```
git clone https://github.com/mbuckler/youtube-bb.git
python3 download.py ./dataset 12
```

### Download pretrained model on VID with 690000 image pairs
Pretrained model is available here [BaiduYun](https://pan.baidu.com/s/14vzkwTJWqCKCAcH-j5WdLg)

### Training Phase 
```
git clone https://github.com/songdejia/siamese-RPN
cd code_v1.0
python train_siamrpn.py --dataroot=/PATH/TO/YOUR/DATASET --lr=0.001 --checkpoint_path=/PATH/TO/YOUR/WEIGHT
```

## Visualization for debug

**bbox in detection**  
green -- ground truth which is got by pos anchor shift with reg_target  
red   -- bbox which is got by pos anchor with reg_pred  
black -- bbox with highest score

<div align=center><img width="400" height="400" src="https://github.com/songdejia/siamese-RPN/blob/master/screenshot/bbox_in_detection.jpg"/></div>


**proposal in original image**
<div align=center><img width="640" height="360" src="https://github.com/songdejia/siamese-RPN/blob/master/screenshot/bbox_in_origin.jpg"/></div>


## Authors  
* **Bo Li** - *paper* - [Siamese-RPN](http://openaccess.thecvf.com/content_cvpr_2018/papers/Li_High_Performance_Visual_CVPR_2018_paper.pdf)
* **De jiasong** - *code* - [Siamese-RPN-pytorch](https://github.com/songdejia/siamese-RPN)
* **Makalo**     - *code* - [Siamese-RPN-tensorflow](https://github.com/makalo/Siamese-RPN-tensorflow.git)