Hello,
I would like to get a sample of pictures showing the various illuminations and conditions. You have mentioned you used opencv face detection, I believe this is Haar Method. I understand you need to change that into more accurate algorithms (in terms of bounding box coordinates)
I also have few more questions:
Is there specific algorithm in mind?
Is it okay to use deep learning based models?
Thanks in advance,
Regards,
Bassel Safwat