Can GhostFaceNets be fine tuned for facial landmark detection?

[m-a-x-c]

I am novice and would appreciate some advice. I have a bunch of faces with custom landmarks (x and y coordinates). I want to fine tune GhostFaceNets to detect these x and y coordinates. My question are:

• Is this possible and will it be difficult?
• Would you recommend that I try doing it with a different model instead of GhostFaceNets?

Any help is deeply appreciated.

[HamadYA]

I am novice and would appreciate some advice. I have a bunch of faces with custom landmarks (x and y coordinates). I want to fine tune GhostFaceNets to detect these x and y coordinates. My question are:

• Is this possible and will it be difficult?
• Would you recommend that I try doing it with a different model instead of GhostFaceNets?

Any help is deeply appreciated.

Hi, GhostFaceNets is a face recognition model, that is built for recognition, verification, or feature extraction purposes to identify people.

To use it as a face detector, or object detector, you need to change the recognition head to a detection head such as YOLO, or DETR and train it on the corresponding data (e.g., WiderFACE for face detection).

Check this github link: https://github.com/open-mmlab/mmdetection

[HamadYA]

Let me know if you have any other questions

I am novice and would appreciate some advice. I have a bunch of faces with custom landmarks (x and y coordinates). I want to fine tune GhostFaceNets to detect these x and y coordinates. My question are:

• Is this possible and will it be difficult?
• Would you recommend that I try doing it with a different model instead of GhostFaceNets?

Any help is deeply appreciated.

[m-a-x-c]

Thanks for the information. I will look into MMDetection.