Exploring the Spectrum of Visio-Linguistic Compositionality and Recognition

TL;DR

This paper evaluates vision-language models like CLIP across multiple benchmarks to understand the trade-offs between compositional understanding and recognition accuracy, highlighting the need for balanced model development.

Contribution

It provides a comprehensive evaluation of existing VLMs on both recognition and compositionality benchmarks, revealing patterns and trade-offs in their capabilities.

Findings

Identified trade-offs between compositionality and recognition accuracy.

Analyzed 274 CLIP checkpoints to reveal performance patterns.

Highlighted the need for balanced models and benchmark formulation.

Abstract

Vision and language models (VLMs) such as CLIP have showcased remarkable zero-shot recognition abilities yet face challenges in visio-linguistic compositionality, particularly in linguistic comprehension and fine-grained image-text alignment. This paper explores the intricate relationship between compositionality and recognition -- two pivotal aspects of VLM capability. We conduct a comprehensive evaluation of existing VLMs, covering both pre-training approaches aimed at recognition and the fine-tuning methods designed to improve compositionality. Our evaluation employs 12 benchmarks for compositionality, along with 21 zero-shot classification and two retrieval benchmarks for recognition. In our analysis from 274 CLIP model checkpoints, we reveal patterns and trade-offs that emerge between compositional understanding and recognition accuracy. Ultimately, this necessitates strategic efforts towards developing models that improve both capabilities, as well as the meticulous formulation of benchmarks for compositionality. We open our evaluation framework at https://github.com/ytaek-oh/vl_compo.