Project page: https://taotaoorange.github.io/projects/SINT/SINT_proj.html
If you find our work useful, please consider citing:
@inproceedings{tao2016sint,
title={Siamese Instance Search for Tracking},
author={Tao, Ran and Gavves, Efstratios and Smeulders, Arnold W M},
booktitle={CVPR},
year={2016}
}
Prerequisites: Fast-RCNN https://github.com/rbgirshick/fast-rcnn (The region-of-interest pooling layer is used in our implementation) or Caffe (with the region-of-interest pooling layer added).
Trained caffe model: http://isis-data.science.uva.nl/rantao/SINT_similarity.caffemodel
[Run the tracker] After adding the normalization layer into your Fast-RCNN or Caffe, and modifying the path where necessary, you should be able to reproduce our experiments on OTB benchmark with the provided model by running 'run_SINT_OTB.py'.
[Training data] The scripts in 'scripts_prepare_training_data' shows how to process ALOV (http://alov300pp.joomlafree.it/dataset-resources.html) to prepare training data to train the Siamese network.
You can find more information, such as the result files on OTB, on the project page https://taotaoorange.github.io/projects/SINT/SINT_proj.html