MER-CLIP uses a frozen CLIP text encoder as a label encoder and a cross-modal decoder to align video, audio, and text features, setting new state-of-the-art results on CMU-MOSI and CMU-MOSEI.
A survey of state-of-the-art approaches for emotion recognition in text
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Leveraging CLIP Encoder for Multimodal Emotion Recognition
MER-CLIP uses a frozen CLIP text encoder as a label encoder and a cross-modal decoder to align video, audio, and text features, setting new state-of-the-art results on CMU-MOSI and CMU-MOSEI.