Meerkat: Audio-Visual Large Language Model for Grounding in Space and Time

TL;DR

Meerkat is a novel audio-visual large language model that achieves fine-grained spatial and temporal understanding of images and audio, enabling it to perform complex grounding and localization tasks with state-of-the-art accuracy.

Contribution

The paper introduces Meerkat, a new multi-modal LLM with a novel modality alignment module and a large curated dataset, advancing fine-grained audio-visual understanding.

Findings

State-of-the-art performance on multiple audio-visual tasks

Up to 37.12% relative improvement over previous methods

Effective spatial and temporal grounding in audio-visual data

Abstract

Leveraging Large Language Models' remarkable proficiency in text-based tasks, recent works on Multi-modal LLMs (MLLMs) extend them to other modalities like vision and audio. However, the progress in these directions has been mostly focused on tasks that only require a coarse-grained understanding of the audio-visual semantics. We present Meerkat, an audio-visual LLM equipped with a fine-grained understanding of image and audio both spatially and temporally. With a new modality alignment module based on optimal transport and a cross-attention module that enforces audio-visual consistency, Meerkat can tackle challenging tasks such as audio referred image grounding, image guided audio temporal localization, and audio-visual fact-checking. Moreover, we carefully curate a large dataset AVFIT that comprises 3M instruction tuning samples collected from open-source datasets, and introduce MeerkatBench that unifies five challenging audio-visual tasks. We achieve state-of-the-art performance on all these downstream tasks with a relative improvement of up to 37.12%.