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
751 stars 180 forks source link

The principal of selecting parameters. #25

Closed iminfine closed 5 years ago

iminfine commented 5 years ago

I notice that different parameters are used when test proposed models with different benchmarks. For example, the parameters of SiamFCRes22 of OTB2015 are scale_step: 1.1897 scale_lr: 0.2226 scale_penalty: 0.9370 w_influence: 0.2897 Could you please detail the selection strategy?

JudasDie commented 5 years ago

I notice that different parameters are used when test proposed models with different benchmarks. For example, the parameters of SiamFCRes22 of OTB2015 are scale_step: 1.1897 scale_lr: 0.2226 scale_penalty: 0.9370 w_influence: 0.2897 Could you please detail the selection strategy?

See tune_tpe.py for details.

iminfine commented 5 years ago

I notice that different parameters are used when test proposed models with different benchmarks. For example, the parameters of SiamFCRes22 of OTB2015 are scale_step: 1.1897 scale_lr: 0.2226 scale_penalty: 0.9370 w_influence: 0.2897 Could you please detail the selection strategy?

See tune_tpe.py for details.

Thank you for your prompt reply!