🎓 CS Ph.D. Candidate @ SMU | 🎬 Video MLLMs | 🤖 Multimodal Agents
Advised by Prof. Lizi Liao at Singapore Management University. Previously a research intern at BAAI with Prof. Zheng Liu and Prof. Bo Zhao.
- 🎬 UniVA - Plan, generate, and edit videos with multimodal agents. Paper
- 🔎 Video-Browser - An agent that searches and watches videos to answer questions. Paper
- 📽️ Video-XL-2 (co-author) - Efficient long-video understanding through task-aware KV sparsification. Paper
- 👁️ MMVU (co-author) - Benchmarking MLLM robustness to misleading questions. Paper
- 📝 Video2Article - From videos to illustrated articles, with agent-selected frames.
- 🖥️ OmniMate - A screen-aware assistant prototype with multimodal memory and MCP tools.
- 📬 ChatEmail - AI email summaries, priority analysis, and an inbox interface.
- 🎮 Escape from Ivory Towerkov - A research-themed extraction game with an academic career loop.
- 🕸️ CrawURLKnowledge - Turn documentation websites into searchable RAG knowledge bases.
- 🎙️ Pod2Article - YouTube search and timestamped transcripts for agent-written articles.
- 💬 ConflictLens - Bilingual communication practice with AI reflection and feedback.
Homepage · Google Scholar · chr1ce [at] foxmail.com