* Take all images in folder and send them to Face - Detect
* Save all given FaceIds with all atributes and save it into local MSSQL database. Also save image name.
* Take all FaceIds from db with Blur-level <> High and send them to Face - Group (max. 1000 FaceIds in one call)
* Set random name to each given group of FaceIds and update every FaceId in db.
* Take FaceId with the lowest Blur level value from each group, merge them by 10 and send them to Face - Identify, against KnownPersonGroup. (Chceck wether PersonGroup is Trained, if not Train it)
* If there is known person (confidence at least 0.5) add this face to this person and train Group. Update FaceId in db - add PersonGroup, PersonId and new FaceId. (Each person in PersonGroup could have no more than 248 faces - need to take control abou it)
* If there is not known person, do previous process againts UnknownPersonGroup. If there is not find person, create a new one.
Hello, Sir. I read your project spec and understood your requirements. As an ML, DL, CV, Image processing Expert, I have rich experiences with Face Detection/Recognition and Classification, similarity retrieval, etc Mi Daha Fazla