Shanghua Gao, a researcher working on computer vision and representation learning. I am a great fan of open source, as it makes research work valuable for many people.
Some of open source projects that I've been working on/contributing to:
- https://github.com/mims-harvard/TxAgent TxAgent: An AI Agent for Therapeutic Reasoning Across a Universe of Tools （TxAgent = Multi-step reasoning + function calls to 200+ tools）
- https://github.com/mims-harvard/ToolUniverse ToolUniverse is a scientific environment of over 600+ tools designed for AI agents.
- https://github.com/mims-harvard/UniTS (UniTS is a unified time series model that can process various tasks across multiple domains with shared parameters.)
- https://github.com/sail-sg/EditAnything (Aims to Edit and Generate Anything in an image with the help of Segment Anything etc.)
- https://github.com/zhoudaquan/ChatAnything (Facetime Chat with LLM-Enhanced Personas.)
- https://github.com/sail-sg/MDT (Masked Diffusion Transformer is the SOTA for image synthesis.)
- https://github.com/sail-sg/TEC (A very fast and effective self-supervised pretraining method towards sustainable self-supervised learning.)
- https://github.com/LUSSeg (The first method and benchmark for large-scale unsupervised semantic segmentation.) ImageNet-S, PASS, ImageNetSegModel, LUSS-mmsegmentation
- https://github.com/Res2Net (A widely used multi-scale backbone network.) Res2Net-PretrainedModels, Res2Net-Pose-Estimation, Res2Net-mmdetection, Res2Net-detectron2, Res2Net-ImageNet-Training, Res2Net-maskrcnn, Res2Net-CSNet, Res2Net-PoolNet
- https://github.com/ShangHua-Gao/SOD100K (Highly efficient salient object detection model with only about 100k parameters.)
- https://github.com/ShangHua-Gao/RFNext (Effective and fast receptive field searching method.)
- https://github.com/ShangHua-Gao/RBN (An improvement of BatchNorm with stronger inference performance.)
- mmdetection mmsegmentation mmcv (Res2Net, RFNext, LUSS are inside openmmlab.)