Results

We adopt the same evaluation metric employed in the PASCAL VOC dataset. If the ratio of the intersection of a detected region with an annotated face region is greater than 0.5, a score of 1 is assigned to the detected region, and 0 otherwise.Further details for the evaluation procedure can be found in our paper.

Evaluation Code

Evaluation and Plotting scripts for generating the PR curves are provided below.

The latest results may not be updated always in the downloaded tar file. In that case, please copy the latest PR curves individually and re-run the plotting scripts.

Reporting results

For questions and result submission, please contact Shuo Yang at shuoyang.1213@gmail.com

Benchmark

For details on the evaluation scheme please see our paper.

Scenario-Ext: A face detector is trained using any external data, and tested on the WIDER FACE test partition.

Performance curves


Scenario-Int: A face detector is trained using WIDER FACE training/validation partitions, and tested on WIDER FACE test partition.

Performance curves for the validation set
Performance curves for the test set