Video Indexer detects less people than Face API, why

Hi Video Indexer team, I have been playing with Video Indexer and Face API. It looks like Video Indexer is detecting “less” people than Face API. For example I indexed a small video using Video Indexer and no people were detected. Then I processed some frames, of the same video, using the Face API and people were detected. What’s the reason behind this? I guess it could be something like: - Video Indexer is not processing each individual frame? - Video Indexer is using a different AI model to detect faces/people? - Or ? Is there more technical details about the way Video Indexer is analyzing the video’s? (more information than https://docs.microsoft.com/en-us/azure/media-services/video-indexer/video-indexer-overview)? Ps. I am talking about detecting, so not even about identification. Regards, Johan.

Comments

  •  
    Hi Johan, VideoIndexer is not using the exact model that Face API uses and there are additional differences in the detection phase. If you want you can send us an example video and we will look into why it doesn't detect faces in it.
    Posted by Hidden Fri, 03 Jul 2020 17:17:51 GMT


You're not signed in. Please sign-in to report an issue or post a comment.