AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Cross-Modal Guided Visual Synthesis for Data-Efficient Multimodal Depression Recognition

Shanliang Yang, Xiaoxiao Wang

CVPR 2026 · Vision-language and multimodal › Cross-modal learning · approach: Audio-visual speech separation

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add cross-modal guided visual feature synthesis to the recognizer (ablation)Multimodal recognizer · DAIC-WOZ classificationF1-Score: 0.825 -> 0.860Improved
add task-guided feedback (augmented loss gradient into synthesizer) versus two-stage decoupled synthesis (ablation)CMG-VS · DAIC-WOZ classificationF1-Score: 0.841 -> 0.860Improved
add two-stage synthesis without task guidance (ablation)Recognizer-Only · DAIC-WOZ classificationF1-Score: 0.825 -> 0.841Improved
condition synthesis on both audio and text instead of text only (ablation)Conditional visual synthesizer · DAIC-WOZ classificationF1-Score: 0.851 -> 0.860Improved
condition synthesis on both audio and text instead of audio only (ablation)Conditional visual synthesizer · DAIC-WOZ classificationF1-Score: 0.845 -> 0.860Improved
synthesize all visual features rather than only AUs (ablation)Conditional visual synthesizer · DAIC-WOZ classificationF1-Score: 0.857 -> 0.860No clear effect
synthesize all visual features rather than only pose and gaze (ablation)Conditional visual synthesizer · DAIC-WOZ classificationF1-Score: 0.837 -> 0.860Improved
add visual synthesis to the recognizer (confusion matrix) (ablation)Multimodal recognizer · DAIC-WOZ test setfalse positives: 5 -> 2 false positives; 13 -> 12 depressed identifiedImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy