ExpAlign: Expectation-Guided Vision-Language Alignment for Open-Vocabulary Grounding

TL;DR

ExpAlign introduces a theoretically grounded, expectation-guided vision-language alignment framework that enhances open-vocabulary detection and segmentation without requiring explicit supervision, achieving state-of-the-art results.

Contribution

The paper presents ExpAlign, a novel alignment framework using multiple instance learning and energy-based regularization for improved weakly supervised vision-language tasks.

Findings

Achieves 36.2 AP_r on LVIS minival, outperforming comparable methods.

Improves open-vocabulary detection and zero-shot segmentation, especially on long-tail categories.

Remains lightweight and inference-efficient.

Abstract

Open-vocabulary grounding requires accurate vision-language alignment under weak supervision, yet existing methods either rely on global sentence embeddings that lack fine-grained expressiveness or introduce token-level alignment with explicit supervision or heavy cross-attention designs. We propose ExpAlign, a theoretically grounded vision-language alignment framework built on a principled multiple instance learning formulation. ExpAlign introduces an Expectation Alignment Head that performs attention-based soft MIL pooling over token-region similarities, enabling implicit token and instance selection without additional annotations. To further stabilize alignment learning, we develop an energy-based multi-scale consistency regularization scheme, including a Top-K multi-positive contrastive objective and a Geometry-Aware Consistency Objective derived from a Lagrangian-constrained free-energy minimization. Extensive experiments show that ExpAlign consistently improves open-vocabulary detection and zero-shot instance segmentation, particularly on long-tail categories. Most notably, it achieves 36.2 AP$_r$ on the LVIS minival split, outperforming other state-of-the-art methods at comparable model scale, while remaining lightweight and inference-efficient.