SJTU Course CS348: Computer Vision (2022 Fall) Team B Project.

• Code in folder improved_diffusion is forked from openai / improved-diffusion under the MIT license.
• Human pose data format refers to HumanData class and tools from open-mmlab/mmhuman3d
  • Model execution relies on MMHuman3d, read the submodule documents for the installation tutorial
• Our diffusion model input data structure is converted from mmhuman3d.HumanData['smpl']
• Group meeting notes and other materials are uploaded to the utility branch