Rethinking Diversified and Discriminative Proposal Generation for Visual Grounding

TLDR

Visual grounding seeks to locate an image region described by a textual query, yet most methods overlook the crucial proposal generation step, treating it as a secondary module. The study reexamines the characteristics that constitute an effective proposal generator for visual grounding. They propose the Diversified and Discriminative Proposal Networks (DDPN) that jointly generate diverse and discriminative proposals, and use these to build a high‑performance visual grounding baseline evaluated on four benchmarks. The approach achieves substantial gains, improving performance by 18.8% on ReferItGame and 8.2% on Flickr30k Entities relative to state‑of‑the‑art methods.

Abstract

Visual grounding aims to localize an object in an image referred to by a textual query phrase. Various visual grounding approaches have been proposed, and the problem can be modularized into a general framework: proposal generation, multi-modal feature representation, and proposal ranking. Of these three modules, most existing approaches focus on the latter two, with the importance of proposal generation generally neglected. In this paper, we rethink the problem of what properties make a good proposal generator. We introduce the diversity and discrimination simultaneously when generating proposals, and in doing so propose Diversified and Discriminative Proposal Networks model (DDPN). Based on the proposals generated by DDPN, we propose a high performance baseline model for visual grounding and evaluate it on four benchmark datasets. Experimental results demonstrate that our model delivers significant improvements on all the tested data-sets (e.g., 18.8% improvement on ReferItGame and 8.2% improvement on Flickr30k Entities over the existing state-of-the-arts respectively).