A Simple Recipe for Contrastively Pre-training Video-First Encoders Beyond 16 Frames
Understanding long, real-world videos requires modeling of long-range visual dependencies. To this end, we explore video-first architectures, building on the common paradigm of transferring large-scale, image--text models to video via shallow temporal fusion. However, we expose two limitations to th...
Gespeichert in:
Hauptverfasser: | , , , , , , , , , |
---|---|
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | Understanding long, real-world videos requires modeling of long-range visual
dependencies. To this end, we explore video-first architectures, building on
the common paradigm of transferring large-scale, image--text models to video
via shallow temporal fusion. However, we expose two limitations to the
approach: (1) decreased spatial capabilities, likely due to poor
video--language alignment in standard video datasets, and (2) higher memory
consumption, bottlenecking the number of frames that can be processed. To
mitigate the memory bottleneck, we systematically analyze the memory/accuracy
trade-off of various efficient methods: factorized attention,
parameter-efficient image-to-video adaptation, input masking, and
multi-resolution patchification. Surprisingly, simply masking large portions of
the video (up to 75%) during contrastive pre-training proves to be one of the
most robust ways to scale encoders to videos up to 4.3 minutes at 1 FPS. Our
simple approach for training long video-to-text models, which scales to 1B
parameters, does not add new architectural complexity and is able to outperform
the popular paradigm of using much larger LLMs as an information aggregator
over segment-based information on benchmarks with long-range temporal
dependencies (YouCook2, EgoSchema). |
---|---|
DOI: | 10.48550/arxiv.2312.07395 |