biubug6 / Pytorch_Retinaface

Retinaface get 80.99% in widerface hard val using mobilenet0.25.
MIT License
2.63k stars 774 forks source link

About preparing an own data for training! #174

Open muxiddin19 opened 3 years ago

muxiddin19 commented 3 years ago

Thank you for your very useful tool. I wanted to use a retina face for my project related to face detection from the CCTV data after anonymizing them. But detection part is failing with some people in the given CCTV data. I wanted to improve the tool's performance and prepared the data from the frames where the model was failed. For the Annotation task, I used some familiar tools like Labelme, AnnoMage, and Make Sense. But it was failed when I trained the model with my prepared data because of the label.txt file. The obtained errors were related to converting string to float and with data shape. Please, can you share the tool which you used for preparing the data for training? I would appreciate it if you do this, which is essential for handling the given project. Thanks in advance. And good luck in your future works.