Junlin Hou

semantic segmentation (1) video captioning (1) multimodal learning (1) gaussian mixture model (1) zero-shot transfer (1) retrieval-augmented generation (1) image-text alignment (1) egocentric video (1) cross-attention module (1) text feature (1) class activation mapping (1) foreground segmentation (1) retrieval-augmented learning (1) weakly supervised object localization (1) open-vocabulary learning (1) concept detection (1) interpretable explanation (1) medical image diagnosis (1) cross-view retrieval (1) masked entity completion (1)