Publications
-
Planning in 8 Tokens: A Compact Discrete Tokenizer for Latent World Model
-
Exploring State-Space Models for Data-Specific Neural Representations
-
Improving Target Presence and Plurality Recognition for Generalized Referring Image Segmentation
-
Planning in 16 Tokens: A Compact Discrete Tokenizer for Latent World Model
-
Classification Matters: Improving Video Action Detection with Class-Specific Attention
-
Detector-Free Weakly Supervised Group Activity Recognition