Modulating early visual processing by language
This work addresses a fundamental problem in multimodal AI by challenging the independent processing paradigm, potentially benefiting researchers in vision-language integration.
The paper tackled the assumption that language does not affect low-level visual processing by proposing to modulate the entire visual processing with linguistic input, resulting in significant improvements on two visual question answering tasks.
It is commonly assumed that language refers to high-level visual concepts while leaving low-level visual processing unaffected. This view dominates the current literature in computational models for language-vision tasks, where visual and linguistic input are mostly processed independently before being fused into a single representation. In this paper, we deviate from this classic pipeline and propose to modulate the \emph{entire visual processing} by linguistic input. Specifically, we condition the batch normalization parameters of a pretrained residual network (ResNet) on a language embedding. This approach, which we call MOdulated RESnet (\MRN), significantly improves strong baselines on two visual question answering tasks. Our ablation study shows that modulating from the early stages of the visual processing is beneficial.