Closed bit-scientist closed 1 year ago
I have a similar problem with the tracker. I just want to use pure YOLOX-X model with 80 classes of COCO for car tracking. But I cannot get the results that I expected. Did you find any solution about it?
So do I, I am trying to find some method for detecting car, but the project is too complex, maybe the team can recommend a part to get start My starting point: exps/example/mot/yolox_s_mix_det.py, is that work? I tried to modify num_classes, but the case don't match
Hi, @YamacTan (sorry for the late reply to your message) and @Everloom-129. I can hardly see an activity on this repo. Good news is that ultralytics has integrated bytetrack into their framework. There you can use bytetrack tracking with any of your custom trained v8 model. Try that and good luck to both of you!
Ohhh, I have just realized it yesterday with official help! That's wonderful, thank you for the fantastic model
Get Outlook for Androidhttps://aka.ms/AAb9ysg
From: bit-scientist @.> Sent: Friday, August 25, 2023 6:15:23 PM To: ifzhang/ByteTrack @.> Cc: Wang, Jie @.>; Mention @.> Subject: Re: [ifzhang/ByteTrack] Do standard models only support person class? (Issue #207)
You don't often get email from @.*** Learn why this is importanthttps://aka.ms/LearnAboutSenderIdentification
Hi, @YamacTanhttps://github.com/YamacTan (sorry for the late reply to your message) and @Everloom-129https://github.com/Everloom-129. I can hardly see an activity on this repo. Good news is that ultralytics https://github.com/ultralytics/ultralytics has integrated bytetrack https://docs.ultralytics.com/modes/track/ into their framework. There you can use bytetrack tracking with any of your custom trained v8 model. Try that and good luck to both of you!
— Reply to this email directly, view it on GitHubhttps://github.com/ifzhang/ByteTrack/issues/207#issuecomment-1693127738, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ARIDAIQH5QTIEUYX2O37U5LXXB3LXANCNFSM5ZLTVWOQ. You are receiving this because you were mentioned.Message ID: @.***>
First off, thank you for this amazing project. It's really promising.
I have some questions, though. 😄 It looks like standard models were trained for only person class, right? If I want to track pet animals (cats and dogs) that exist in COCO dataset, can I just use the custom-pretrained YOLOX detector (cat and dog)? You mentioned that one needs to modify byte_tracker.py if s(h)e wants to combine other detectors with BYTE. Could you elaborate what exactly needs to be done? I finding it a little complicated to start. THANK YOU!