Brad Dwyer
Brad Dwyer
We don't have anyone actively maintaining or updating this; have been using ByteTrack more internally. PRs welcome though!
(Leaving open until resolved & merged)
I don't know if this is still bugging anyone else but after 3 years I finally couldn't stand it anymore. There's been a PR open here for 2 years that...
Hi, we will likely be able to fix this as a stopgap but please note that [Jetpack 4 is EOL](https://forums.developer.nvidia.com/t/jetpack-4-reaches-end-of-life/267563) and no longer supported by NVIDIA. Please consider upgrading. We...
Hi @Qfc9, [the default](https://github.com/roboflow/inference/blob/17adaedf813ed944a12f2951547d2b4284dda680/inference/core/env.py#L235) is to use the CUDA execution provider, then fallback to OpenVINO, then fallback to CPU if those are not available. This can be overridden by setting...
Definitely agree our docs are lacking here. Would definitely accept a PR to make this more clear. Having two competing ways to do set these values seems not ideal. Via...
Should we add torch as a dep for the tests?
> a way to display in UI that this block wont work unless you run inference locally We'll certainly need this ability for other blocks (even if we eventually host...
Also having this issue.
Could you also include a benchmark to see if this improves speed? I have had mixed results with MPS (it even degrades performance in some cases depending on how few...