OmniVL:One Foundation Model for Image-Language and Video-Language Tasks

TL;DR

OmniVL introduces a unified transformer-based foundation model capable of handling both image-language and video-language tasks through joint pretraining and a novel contrastive loss, achieving state-of-the-art results across diverse tasks.

Contribution

The paper proposes a universal architecture for image and video-language tasks, utilizing decoupled joint pretraining and a unified contrastive loss to enhance multi-modal understanding.

Findings

Supports a wide range of tasks without task-specific adaptors

Achieves state-of-the-art or competitive results on multiple benchmarks

Effectively leverages both supervised and noisily supervised data

Abstract

This paper presents OmniVL, a new foundation model to support both image-language and video-language tasks using one universal architecture. It adopts a unified transformer-based visual encoder for both image and video inputs, and thus can perform joint image-language and video-language pretraining. We demonstrate, for the first time, such a paradigm benefits both image and video tasks, as opposed to the conventional one-directional transfer (e.g., use image-language to help video-language). To this end, we propose a decoupled joint pretraining of image-language and video-language to effectively decompose the vision-language modeling into spatial and temporal dimensions and obtain performance boost on both image and video tasks. Moreover, we introduce a novel unified vision-language contrastive (UniVLC) loss to leverage image-text, video-text, image-label (e.g., image classification), video-label (e.g., video action recognition) data together, so that both supervised and noisily supervised pretraining data are utilized as much as possible. Without incurring extra task-specific adaptors, OmniVL can simultaneously support visual only tasks (e.g., image classification, video action recognition), cross-modal alignment tasks (e.g., image/video-text retrieval), and multi-modal understanding and generation tasks (e.g., image/video question answering, captioning). We evaluate OmniVL on a wide range of downstream tasks and achieve state-of-the-art or competitive results with similar model size and data scale.