siamese_rpn

Siamrpn++: Evolution of Siamese Visual Tracking With Very Deep Networks

Introduction

[ALGORITHM]

@inproceedings{li2019siamrpn++,
  title={Siamrpn++: Evolution of siamese visual tracking with very deep networks},
  author={Li, Bo and Wu, Wei and Wang, Qiang and Zhang, Fangyi and Xing, Junliang and Yan, Junjie},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={4282--4291},
  year={2019}
}

Results and models

LaSOT

Note that the checkpoints from 10-th to 20-th epoch will be evaluated during training. You can find the best checkpoint from the log file.

We observe around 1.0 points fluctuations in Success and 1.5 points fluctuations in Norm percision. We provide the best model with its configuration and training log.

Backbone	Style	Lr schd	Mem (GB)	Inf time (fps)	Success	Norm precision	Config	Download
R-50	-	20e	7.54	50.0	49.9	57.9	config	model | log

UAV123

The checkpoints from 10-th to 20-th epoch will be evaluated during training. You can find the best checkpoint from the log file.

If you want to get better results, you can use the best checkpoint to search the hyperparameters on UAV123 following here. Experimentally, the hyperparameters search on UAV123 can bring around 1.0 Success gain.

The results below are achieved without hyperparameters search. We observe less than 0.5 points fluctuations both in Success and Percision.

Backbone	Style	Lr schd	Mem (GB)	Inf time (fps)	Success	Norm Precision	Precision	Config	Download
R-50	-	20e	7.54	-	60.6	76.5	80.5	config	model | log

TrackingNet

The results of SiameseRPN++ in TrackingNet are reimplemented by ourselves. The best model on LaSOT is submitted to the evaluation server on TrackingNet Chanllenge. We observe less than 0.5 points fluctuations both in Success and Percision. We provide the best model with its configuration and training log.

Backbone	Style	Lr schd	Mem (GB)	Inf time (fps)	Success	Norm precision	Precision	Config	Download
R-50	-	20e	7.54	-	70.6	77.6	65.7	config	model | log

OTB100

The checkpoints from 10-th to 20-th epoch will be evaluated during training. You can find the best checkpoint from the log file.

If you want to get better results, you can use the best checkpoint to search the hyperparameters on OTB100 following here. Experimentally, the hyperparameters search on OTB100 can bring around 1.0 Success gain.

Note: We train the SiameseRPN++ in the official pysot codebase and can not reproduce the same results reported in the paper. We only get 66.1 Success and 86.7 Precision by following the training and hyperparameters searching instructions of pysot, which are lower than those of the paper by 3.5 Succuess and 4.7 Precision respectively. In our codebase, the Success and Precision are lower 4.8 and 3.7 respectively than those of the paper. Notably, the results below are achieved without hyperparameters search. We observe around 0.5 points fluctuations both in Success and Percision.

Backbone	Style	Lr schd	Mem (GB)	Inf time (fps)	Success	Norm Precision	Precision	Config	Download
R-50	-	20e	-	-	64.8	83.2	87.7	config	model | log