My project is to do real-time face detection and apply filters over it like MSQRD/Snapchat. I am new to this field so I need a road map on how to do it. I am thinking of referring this paper. I know basic ML and have a math background.
What I had done till now:
I have read old answers on this topic; first, second and third. The second one is good, but it is inclined towards mobile applications and others didn't help much. Also, the libraries which they suggested were paid.
I did some googling and found that some tutorials are referring to this as Augmented Reality and some this as a GAN network. For human face detection, most of the tutorials are using Viola-Jones Algorithm. Face detection can be achieved easily, but the hard part is to do face alignment. From what I had seen in tutorials, First, capture an image and applied the viola jones algorithm to it, and train the model to fit the face more accurately. Then convert that 2d image into a 3d mesh(I don't know how). Afterward, apply the filter as an overlay above the face.
I have half-baked knowledge of the subject. So, I am kinda confused about what things to study and in which order. It would be really helpful if you guys can give some insights.
Thank You.
