Research artifacts spanning AI4Earth agents, multimodal foundation models, and real-world perception datasets. Each card prioritizes what exists, what I contributed, and where to try or inspect it.
Self-evolving AI agent for climate-science workflows.
Agent workflow design, climate-science reasoning interface, and system evaluation.
Explores climate-science reasoning, tool use, and iterative scientific workflows.
Large-scale, multi-view 3D human pose dataset and benchmark under real-world conditions.
Dataset construction, benchmark design, annotation pipeline, and evaluation.
Provides smartphone-captured multi-view data, evaluation baselines, and real-world transfer analysis for 3D human pose estimation.
Instruction-tuning dataset, framework, and benchmark for multimodal large language models.
Instruction data construction, multimodal benchmark design, and framework development.
Introduces LAMM-Dataset and LAMM-Benchmark for 2D image and 3D point-cloud understanding, plus an extensible framework for modality expansion.