Vid2Seq: Large-Scale Pretraining of a Visual Language Model for Dense Video Captioning

Abstract

In this work, we introduce Vid2Seq, a multi-modal single-stage dense event captioning model pretrained on narrated videos which are readily-available at scale. The Vid2Seq architecture augments a language model with special time tokens, allowing it to seamlessly predict event boundaries and textual descriptions in the same output sequence. Such a unified model requires large-scale training data, which is not available in current annotated datasets. We show that it is possible to leverage unlabeled narrated videos for dense video captioning, by reformulating sentence boundaries of transcribed speech as pseudo event boundaries, and using the transcribed speech sentences as pseudo event captions. The resulting Vid2Seq model pretrained on the YT-Temporal-1B dataset improves the state of the art on a variety of dense video captioning benchmarks including YouCook2, ViTT and ActivityNet Captions. Vid2Seq also generalizes well to the tasks of video paragraph captioning and video clip captioning, and to few-shot settings. Our code is publicly available at https://antoyang.github.io/vid2seq.html.

CVPR 2023 Camera-Ready; Project Webpage: https://antoyang.github.io/vid2seq.html ; 18 pages; 6 figures

References in corpus (9)

• YouTube-8M: A Large-Scale Video Classification Benchmark
• BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation
• Align before Fuse: Vision and Language Representation Learning with Momentum Distillation
• VALUE: A Multi-Task Benchmark for Video-and-Language Understanding Evaluation
• iPerceive: Applying Common-Sense Reasoning to Multi-Modal Dense Video Captioning and Video Question Answering
• Multimodal Pretraining for Dense Video Captioning
• Obj2Seq: Formatting Objects as Sequences with Class Prompt for Visual Tasks
• Contrastive Language-Action Pre-training for Temporal Localization
• Contrastive Video-Language Learning with Fine-grained Frame Sampling