Visual Set Program Synthesizer
This addresses the challenge of compositional visual reasoning for AI assistants, offering a more systematic alternative to black-box models, though it is incremental in applying program synthesis to a specific domain.
The paper tackles the problem of set-based visual reasoning, such as comparing objects in images, by proposing a visual program synthesis approach that generates symbolic programs for execution, and it significantly outperforms state-of-the-art baselines on complex tasks with improved accuracy.
A user pointing their phone at a supermarket shelf and asking "Which soda has the least sugar?" poses a difficult challenge for current visual Al assistants. Such queries require not only object recognition, but explicit set-based reasoning such as filtering, comparison, and aggregation. Standard endto-end MLLMs often fail at these tasks because they lack an explicit mechanism for compositional logic. We propose treating visual reasoning as Visual Program Synthesis, where the model first generates a symbolic program that is executed by a separate engine grounded in visual scenes. We also introduce Set-VQA, a new benchmark designed specifically for evaluating set-based visual reasoning. Experiments show that our approach significantly outperforms state-of-the-art baselines on complex reasoning tasks, producing more systematic and transparent behavior while substantially improving answer accuracy. These results demonstrate that program-driven reasoning provides a principled alternative to black-box visual-language inference.