search

researchmm / SiamDW

[CVPR'19 Oral] Deeper and Wider Siamese Networks for Real-Time Visual Tracking
http://openaccess.thecvf.com/content_CVPR_2019/html/Zhang_Deeper_and_Wider_Siamese_Networks_for_Real-Time_Visual_Tracking_CVPR_2019_paper.html
MIT License
750 stars 180 forks source link

Was the snapshot model trained by VID？ #2

Closed herbert-chen closed 5 years ago

herbert-chen commented 5 years ago

I ran ./snapshot/CIResNet22.pth on OTB2015 and got 0.6437 auc

JudasDie commented 5 years ago

I ran ./snapshot/CIResNet22.pth on OTB2015 and got 0.6437 auc

Please check python version again (including pytorch/opencv/numpy version).

JudasDie commented 5 years ago

I ran ./snapshot/CIResNet22.pth on OTB2015 and got 0.6437 auc

And the provided models for SiamFC are trained on VID.

herbert-chen commented 5 years ago

torch.version '0.3.1' numpy.version '1.12.1' cv2.version '3.1.0'

According to the results you provided(0.644), it is just about right. And could you afford some models trained on GOT, thank you~

JudasDie commented 5 years ago

torch.version '0.3.1' numpy.version '1.12.1' cv2.version '3.1.0'

According to the results you provided(0.644), it is just about right. And could you afford some models trained on GOT, thank you~

I'll provide got10k models later. Thank you for your attention. This question will be closed.