GSR-BENCH: A Benchmark for Grounded Spatial Reasoning Evaluation via Multimodal LLMs

TL;DR

This paper introduces GSR-BENCH, a comprehensive benchmark for evaluating the spatial reasoning abilities of multimodal large language models across various sizes and training methods, revealing their strengths and weaknesses.

Contribution

It extends the What'sUp dataset and provides a detailed evaluation of 27 models, including multimodal LLMs, to analyze their spatial reasoning performance and scaling laws.

Findings

Multimodal LLMs show varying spatial reasoning capabilities.

Model size and training methods influence spatial understanding.

Benchmark highlights specific strengths and weaknesses of models.

Abstract

The ability to understand and reason about spatial relationships between objects in images is an important component of visual reasoning. This skill rests on the ability to recognize and localize objects of interest and determine their spatial relation. Early vision and language models (VLMs) have been shown to struggle to recognize spatial relations. We extend the previously released What'sUp dataset and propose a novel comprehensive evaluation for spatial relationship understanding that highlights the strengths and weaknesses of 27 different models. In addition to the VLMs evaluated in What'sUp, our extensive evaluation encompasses 3 classes of Multimodal LLMs (MLLMs) that vary in their parameter sizes (ranging from 7B to 110B), training/instruction-tuning methods, and visual resolution to benchmark their performances and scrutinize the scaling laws in this task.