Skip to content

IMPLabUniPr/unbiased-teacher

 
 

Repository files navigation

Improving Localization for Semi-Supervised Object Detection

Installation

See unbiased-teacher official pages to know the installation procedure.

Contributes

To see contributes in diff formats.

Models

Most important files.

Experiments

The following configuration files have been used to run each experiment:

Table 1 / Figure 3a:

Row Beta AP
base 31.027
0 0.5 31.775
1 1 31.947
2 2 31.754
3 4 30.445

Table 2 / Figure 3c:

Row Model AP
1 UT 31.027
2 Ours (with filter) 31.605
3 Ours (w/out filter) 31.509

Table 3 / Figure 3d:

Row Mu AP
1 0.5 31.199
2 0.6 31.128
3 0.7 31.461
4 0.75 31.604
5 0.8 31.336
6 0.9 27.125

Table 4:

Row Theta AP
0 0.3 31.404
1 0.4 31.630
2 0.5 31.604
3 0.6 31.158
4 0.7 30.649

Table 5:

Row Lregunsup xsh scores deltas Model AP
1 model 31.027
2 x model 31.947
3 x x model 31.754
4 x x x model 32.166
5 x x x x model 31.923
6 x x x model 31.630

About

Improving Localization for Semi-Supervised Object Detection

Resources

License

Code of conduct

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%