Mike

Results 228 comments of Mike

I just tried it out, it does happen with Aim as the only logger as well when logging remotely.

I tried this example: https://github.com/aimhubio/aim/blob/main/examples/pytorch_lightning_track.py Adding the `repo` argument in AimLogger as `repo='aim://xx.xx.xx.xx:53800'` where xx.xx.xx.xx is the machine where `aim server` is running

Dropping some ideas here: - Periodic Embeddings: For every n frames, we rely entirely on motion-based tracking algorithms (such as Kalman Filter, optical flow, or other predictive models) to estimate...

Yes, it would be interesting to provide the option to feed a camera configuration file in order to convert the 2D object to 3D. Then it would be possible to...

For BoTSORT this would work the same

Great @jahongir7174! Are you planning to train s, m, l, x as well? :smile_cat:

Alright :smile:. Thanks for letting me know. Is this model trained on visible body only (or head as well) are what was the training setup?

Alright, so visible full body. Thanks again :smile: