Skip to content

Latest commit

 

History

History
34 lines (24 loc) · 3.95 KB

MODEL_ZOO.md

File metadata and controls

34 lines (24 loc) · 3.95 KB

PySOT Model Zoo

Introduction

This file documents a large collection of baselines trained with pysot. All configurations for these baselines are located in the experiments directory. The tables below provide results about inference. Links to the trained models as well as their output are provided.

Visual Tracking Baselines

Short-term Tracking

Model
(arch+backbone+xcorr)
VOT16
(EAO/A/R)
VOT18
(EAO/A/R)
VOT19
(EAO/A/R)
OTB2015
(AUC/Prec.)
VOT18-LT
(F1)
Speed
(fps)
url
siamrpn_alex_dwxcorr 0.393/0.618/0.238 0.352/0.576/0.290 0.260/0.573/0.547 - - 180 link
siamrpn_alex_dwxcorr_otb - - - 0.666/0.876 - 180 link
siamrpn_r50_l234_dwxcorr 0.464/0.642/0.196 0.415/0.601/0.234 0.287/0.595/0.467 - - 35 link
siamrpn_r50_l234_dwxcorr_otb - - - 0.696/0.914 - 35 link
siamrpn_mobilev2_l234_dwxcorr 0.455/0.624/0.214 0.410/0.586/0.229 0.292/0.580/0.446 - - 75 link
siammask_r50_l3 0.455/0.634/0.219 0.423/0.615/0.248 0.283/0.597/0.461 - - 56 link
siamrpn_r50_l234_dwxcorr_lt - - - - 0.629 20 link

The models can also be downloaded from Baidu Yun Extraction Code: j9yb

Note:

  • speed tested on GTX-1080Ti
  • alex denotes AlexNet, r50_lxyz denotes the outputs of stage x, y, and z in ResNet-50, and mobilev2 denotes MobileNetV2.
  • dwxcorr denotes Depth-wise Cross Correlation. See more in SiamRPN++ Section 3.4.
  • The suffixes otb and lt are designed for the OTB and VOT long-term tracking challenge, the default (without suffix) is designed for VOT short-term tracking challenge.
  • All above models are trained on VID,YoutubeBB,COCO,ImageNetDet which are the same as DaSiamRPN.
  • The model of SiamFC is from the author's Matlab implementation. The code for further model converting is available in the corresponding experiment file folders.

License

All models available for download through this document are licensed under the Creative Commons Attribution-ShareAlike 3.0 license.