Multimedia Computing

Multimedia Computing

Multimedia computing focuses on the processing and management of media objects such as image, video, text, and audio. Of particular interest is the effective computing and understanding of different media for novel and interactive multimedia applications. The department has established a strong track record in the broad areas of multimedia computing, ranging from motion analysis, 3D human shape recovery, action prediction, personalized recommendations from multimedia data, handwriting performance determination, information retrieval, big data analytics, etc.

Projects in this area have included: predicting human action from 3D skeletal sequences, 3D human pose and shape reconstruction from video, personalized recommendations based on co-purchase data, body shape-aware outfit completion for full-body portrait images, computerized handwriting speed test system (CHSTS), smart handwriting and recognition platform (SHARP). In particular, the SHARP system has been adopted by occupational therapists to help them screen primary school students who may have specific learning difficulties; and the CHSTS system has been supported by Hong Kong Examinations and Assessment Authority to determine extra time allowance in public examination for secondary school students with special education needs.









  • Human Motion Analysis
    • Human Motion Prediction
    • 3D Human Pose and Shape Recovery from Video
    • Dynamic Multi-Person Synthesis from Sparse Views
  • Personalized Recommendation
    • Hierarchical Visual-Aware Minimax Ranking Based on Co-Purchase Data
    • Body Shape-Aware Object-Level Outfit Completion for Full-Body Portrait Images
  • Handwriting Performance Analysis
    • Computerized Handwriting Speed Test System
    • Smart Handwriting and Recognition Platform

Research Centres