Nicolas Hug

See https://github.com/pytorch/vision/pull/8211 for related investigations. This relates to ffmpeg6 but it's probably relevant for all ffmpeg versions

Hi @mantasu , we've added the main relevant info about TVTensors in this tutorial https://pytorch.org/vision/main/auto_examples/transforms/plot_transforms_getting_started.html#what-are-tvtensors which is linked to as the main entrypoint for all things related to TVTensors /...

Sorry @anirudh6415 it's difficult to help without having a minimal reproducible example. Would you mind sharing exactly the steps that you've been using (but as minimal as possible)?

Thanks for opening this issue @kero-ly . From https://github.com/pytorch/vision/issues/7947#issuecomment-1733310035 it looks like PIL and torchvision provide the same output dimensions (they agree on H and W, despite different conventions) and...

Some great progress with `torch-2.3.0.dev20240117+cpu` and torchvision @ 1de7a74a8b93483f1703eef0b306e0ec68e0cd9d where 72 failing tests were resolved from all of the `dynamic=True` jobs.

Thanks for the PR @atharvas The test failure in https://github.com/pytorch/vision/actions/runs/7394640746/job/20496609120?pr=8194 seems relevant, as `SimpleCopyPaste` internally relies on `masks_to_boxes`. But beyond the correctness issue, I am wondering whether such a change...

Thank you for the details @atharvas . That SGTM. Before merging, do you mind running the benchmark on CUDA as well to make sure there's no slow-down for GPUs? If...

Thank you for the report @Tomsen1410 . @ahmadsharif1 curious if you have encountered this issue in your investigations already?

Hi @Caspeerrr , I'm really sorry but unfortunately, my only recommendation at this point will be to *not* use the cuda video decoder from torchvision. It's still in BETA stage...

Cat: release branch cleanup Release branch PR: https://github.com/pytorch/vision/pull/8152 - merged main PR: NA