Towards Domain-Generalized Open-Vocabulary Object Detection: A Progressive Domain-invariant Cross-modal Alignment Method

TL;DR

This paper identifies the vulnerability of open-vocabulary object detection to domain shifts and proposes a progressive alignment method to enhance cross-modal invariance and robustness.

Contribution

It introduces PICA, a novel curriculum-based training approach that improves domain generalization in open-vocabulary object detection by maintaining stable cross-modal alignment.

Findings

Visual shifts cause collapse of cross-modal space in OVOD.

PICA improves robustness to domain shifts in open-vocabulary detection.

The work offers a new benchmark for evaluating domain generalization in OVOD.

Abstract

Open-Vocabulary Object Detection (OVOD) has achieved remarkable success in generalizing to novel categories. However, this success often rests on the implicit assumption of domain stationarity. In this work, we provide a principled revisit of the OVOD paradigm, uncovering a fundamental vulnerability: the fragile coupling between visual manifolds and textual embeddings when distribution shifts occur. We first systematically formalize Domain-Generalized Open-Vocabulary Object Detection (DG-OVOD). Through empirical analysis, we demonstrate that visual shifts do not merely add noise; they cause a collapse of the latent cross-modal space where novel category visual signals detach from their semantic anchors. Motivated by these insights, we propose Progressive Domain-invariant Cross-modal Alignment (PICA). PICA departs from uniform training by introducing a multi-level ambiguity and signal strength curriculum. It builds adaptive pseudo-word prototypes, refined via sample confidence and visual consistency, to enforce invariant cross-domain modality alignment. Our findings suggest that OVOD's robustness to domain shifts is intrinsically linked to the stability of the latent cross-modal alignment space. Our work provides both a challenging benchmark and a new perspective on building truly generalizable open-vocabulary systems that extend beyond static laboratory conditions.