UniTAB: Unifying Text and Box Outputs for Grounded Vision-Language Modeling

TL;DR

UniTAB unifies text and box outputs in a single model for grounded vision-language tasks, improving interpretability and performance across multiple benchmarks by representing outputs with a shared token sequence.

Contribution

It introduces a novel unified token sequence approach with a special <obj> token, enabling simultaneous text and box generation with alignments, simplifying architecture and enhancing performance.

Findings

Outperforms state-of-the-art in grounded captioning and captioning tasks.

Achieves comparable or better results than task-specific models across 7 VL benchmarks.

Parameter-efficient and generalizable to new tasks.

Abstract

We propose UniTAB that Unifies Text And Box outputs for grounded vision-language (VL) modeling. Grounded VL tasks such as grounded captioning require the model to generate a text description and align predicted words with object regions. To achieve this, models must generate desired text and box outputs together, and meanwhile indicate the alignments between words and boxes. In contrast to existing solutions that use multiple separate modules for different outputs, UniTAB represents both text and box outputs with a shared token sequence, and introduces a special <obj> token to naturally indicate word-box alignments in the sequence. UniTAB thus could provide a more comprehensive and interpretable image description, by freely grounding generated words to object regions. On grounded captioning, UniTAB presents a simpler solution with a single output head, and significantly outperforms state of the art in both grounding and captioning evaluations. On general VL tasks that have different desired output formats (i.e., text, box, or their combination), UniTAB with a single network achieves better or comparable performance than task-specific state of the art. Experiments cover 7 VL benchmarks, including grounded captioning, visual grounding, image captioning, and visual question answering. Furthermore, UniTAB's unified multi-task network and the task-agnostic output sequence design make the model parameter efficient and generalizable to new tasks.