search

mikel-brostrom / yolo_tracking

BoxMOT: pluggable SOTA tracking modules for segmentation, object detection and pose estimation models
GNU Affero General Public License v3.0
6.19k stars 1.65k forks source link

People tracking in videos that switch back and forth between multiple camera positions #1410

Open shiboyang opened 2 weeks ago

shiboyang commented 2 weeks ago

Search before asking

Question

Hi there, I am developing a face recognition software, I want to detect the characters appearing in the video, my idea is to detect the characters appearing in the video through a character tracking algorithm and then filter out the main characters in the video based on the number of times the character appears . My video is some news material, and the camera of the video will switch back and forth. I hope to use yolo_track for character tracking, but I tested the tracking algorithm currently supported by the software, but the effect is not very good, especially when switching lenses will update the character's ID. I have very little experience with character tracking and would love some advice from you.

In the attachment I have uploaded a small example https://github.com/mikel-brostrom/yolo_tracking/assets/44657961/2cd99b45-fca9-41f8-b7f8-7fc39fea469c

shiboyang commented 2 weeks ago

My idea is to detect the person who appears the most in the video through person tracking, then use the face detection and head pose estimation algorithm to obtain the front of the person's face, and finally use the face recognition algorithm to obtain the person's name. I would be very grateful if I could get your help

github-actions[bot] commented 1 day ago

👋 Hello, this issue has been automatically marked as stale because it has not had recent activity. Please note it will be closed if no further activity occurs. Feel free to inform us of any other issues you discover or feature requests that come to mind in the future. Pull Requests (PRs) are also always welcomed!