Visual Grounding with Attention-Driven Constraint Balancing
This addresses the visual grounding problem for computer vision researchers by providing an incremental improvement to transformer-based models through better loss function design.
The paper tackles the problem that existing visual grounding methods use object detection losses that only optimize bounding box regression, failing to fully align visual features with language expressions. The proposed Attention-Driven Constraint Balancing (AttBalance) framework achieves constant improvements across five models on four benchmarks and attains new state-of-the-art performance when integrated into QRNet.
Unlike Object Detection, Visual Grounding task necessitates the detection of an object described by complex free-form language. To simultaneously model such complex semantic and visual representations, recent state-of-the-art studies adopt transformer-based models to fuse features from both modalities, further introducing various modules that modulate visual features to align with the language expressions and eliminate the irrelevant redundant information. However, their loss function, still adopting common Object Detection losses, solely governs the bounding box regression output, failing to fully optimize for the above objectives. To tackle this problem, in this paper, we first analyze the attention mechanisms of transformer-based models. Building upon this, we further propose a novel framework named Attention-Driven Constraint Balancing (AttBalance) to optimize the behavior of visual features within language-relevant regions. Extensive experimental results show that our method brings impressive improvements. Specifically, we achieve constant improvements over five different models evaluated on four different benchmarks. Moreover, we attain a new state-of-the-art performance by integrating our method into QRNet.