The inference results on the video are not marked

megvii-research / MOTR

[ECCV2022] MOTR: End-to-End Multiple-Object Tracking with TRansformer

Other

633 stars 93 forks source link

The inference results on the video are not marked #71

Open mengmimi opened 1 year ago

mengmimi commented 1 year ago

I made a simple inference on demo.avi in the command you gave me, using the weight given in MAIN RESULTS. After running successfully, no detection and tracking marks were found in the video in the output file, which is exactly the same as the original video. Why. I would greatly appreciate it if you could provide an answer.

mengmimi commented 1 year ago

To add, it's not the demo. avi provided, but a video of a car driving

carpeug commented 1 year ago

Hi @mengmimi, I am in the same situation. I've just used the script demo.py on a video recorded on autonomous car. I've used the weights pretrained on BDD100K and I got an empty output. Did you solve it in any way or have you any idea?