Awesome
SiamFC - MXNet
:high_brightness: MXNet/gluon :high_brightness: port of the tracking method described in the paper Fully-Convolutional Siamese nets for object tracking
Running the tracker
- Set
root_dataset
inhyperparams/params.json
to your dataset path - Set
all
fromhyperparams/params.json
totrue
or to set one or more specific sequences (e.g.Basketball
and/orSoccer
) invideo
with settingall
tofalse
- See if you are happy with the default parameters in
hyperparams/params.json
- Call the main script
python run_tracker.py
- The results.mat can be found in
results
which you can run on OTB
Results on test sequence:
AUC (%) on OTB
Tracker | OTB2013 | OTB2015 |
---|---|---|
paper (SiamFC_3s) | 60.8 | 58.2 |
ours (SiamFC_MXnet) | 60.9 | 58.8 |
Precision (%) on OTB
Tracker | OTB2013 | OTB2015 |
---|---|---|
paper (SiamFC_3s) | 80.9 | 77.3 |
ours (SiamFC_MXnet) | 81.4 | 76.7 |
Note
Some errors in fixed_crop
and resize
need to be fixed, which caused me a lot of headaches
References
@inproceedings{bertinetto2016fully,
title={Fully-Convolutional Siamese Networks for Object Tracking},
author={Bertinetto, Luca and Valmadre, Jack and Henriques, Joao F and Vedaldi, Andrea and Torr, Philip H S},
booktitle={ECCV 2016 Workshops},
pages={850--865},
year={2016}
}