Emotion-Qwen reports a dual-expert multimodal architecture and a 40K-clip VER dataset that achieve strong emotion recognition while preserving general vision-language performance.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.MM 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Emotion-Qwen: A Unified Framework for Emotion and Vision Understanding
Emotion-Qwen reports a dual-expert multimodal architecture and a 40K-clip VER dataset that achieve strong emotion recognition while preserving general vision-language performance.