I want to use my face detection data set for face model training. My standard face tool is labelme, which is in json format. I saw that many face detection algorithms are based on widerface for training, so I want to know How do I convert my face frame data into widerface format? Thank you

[goldwater668]

No description provided.

[nixeneko]

Since I don't know about your project, I cannot give a proper answer, I suppose.

I am not familiar with Labelme, but it seems to be able to export in the Pascal VOC format. The VOC format seems to be more popular in object detection tasks. (I can find many WIDER FACE-to-Pascal VOC format coverter, but no opposite.)

As face detection is a special case of object detection, which has only one class ('face' ), so you can use object detection examples based on a self-made Pascal VOC format dataset.

I am hoping your project will success.