search

biubug6 / Face-Detector-1MB-with-landmark

1M人脸检测模型(含关键点)
MIT License
1.02k stars 271 forks source link

widerface亲测疑问 #13

Closed ufoym closed 4 years ago

ufoym commented 4 years ago

先致谢作者，很棒的工作！ 我亲测了RFB在widerface的结果，似乎有所差异：

==================== Results ====================
Easy   Val AP: 0.7425689536559531
Medium Val AP: 0.5656252263367466
Hard   Val AP: 0.24684904010538217
=================================================

我用默认提供的RFB模型，长边320等比缩放，thres_cls = 0.6，thres_nms = 0.4。

测试过程：

  1. 下载官方的WIDER Face Validation Images作为dataset_folder，并运行test_widerface.py得到一堆文本结果；
  2. 下载官方eval_tools.zip里的ground_truth，并运行evaluation.py得到上述结果。

不知我是不是哪里没搞对？

biubug6 commented 4 years ago

把thre_cls=0.02，再测

ufoym commented 4 years ago

Got it.

==================== Results ====================
Easy   Val AP: 0.8152261237745217
Medium Val AP: 0.7100352944190249
Hard   Val AP: 0.36318352172760204
=================================================