A Simple Recipe for Contrastively Pre-Training Video-First Encoders Beyond 16 Frames
Citations Over TimeTop 10% of 2024 papers
Abstract
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).
Related Papers
- Recipe Recommendation Method by Considering the User's Preference and Ingredient Quantity of Target Recipe(2014)
- → IYASHI Recipe: Cooking Recipe Recommendation for Healing based on Physical Conditions and Human Relations(2021)1 cited
- → A Historical Study on the Changes in the Recipe of Naengmyeon (Korean Cold Noodles) Base on Water - Focus on the Recipe Data Published in Korea from 1800’s to 1980’s -(2011)2 cited
- → Batch and Hybrid Processes(2006)2 cited
- → Extraction of Cooking Tips from a Recipe and Application to Another Recipe(2020)