[Multimodal] Add PyAV video backend for concurrent video decoding (#39986)

Signed-off-by: Jaseel Muhammad <jaseel.muhammad@mbzuai.ac.ae>
Signed-off-by: Isotr0py <2037008807@qq.com>
Co-authored-by: Isotr0py <2037008807@qq.com>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
This commit is contained in:
Jaseel Muhammad
2026-04-21 20:14:57 -07:00
committed by GitHub
co-authored by Isotr0py Isotr0py
parent 2463f00fb6
commit 6f2c71be8f
4 changed files with 292 additions and 120 deletions
+4 -3
View File
@@ -829,9 +829,10 @@ environment_variables: dict[str, Callable[[], Any]] = {
"VLLM_MAX_AUDIO_CLIP_FILESIZE_MB": lambda: int(
os.getenv("VLLM_MAX_AUDIO_CLIP_FILESIZE_MB", "25")
),
# Backend for Video IO
# - "opencv": Default backend that uses OpenCV stream buffered backend.
# - "identity": Returns raw video bytes for model processor to handle.
# Backend for Video IO — selects the frame-sampling algorithm.
# - "opencv": uniform sampling.
# - "opencv_dynamic": duration-aware dynamic sampling.
# - "identity": returns raw video bytes for model processor to handle.
#
# Custom backend implementations can be registered
# via `@VIDEO_LOADER_REGISTRY.register("my_custom_video_loader")` and