Vision 汇总(archived)

目录#

• Multimodal *
  • Survey
  • 视觉理解
  • 生成
  • 端到端训练LLM(LMM)
  • Multimodal Agent*
  • Data
• Vision

Multimodal *#

Survey#

视觉理解#

• Vision Encoder

  • multimodal learning
  • Supervised Learning
  • self-distillation
  • Auto-encoding
  • Masked Modeling
    • MAE

• Connector

• Segmentation

生成#

• Diffusion

  • Stable Diffusion

• Controllable

  • Spatial Control

    • Dense control

      • single

      • multi

    • layout/box

      • GLIGEN
      • Reco

  • Style control

    • subject-driven

      • concept customization [fine-tuning]

    • image-driven

  • Sementic Control

  • 其他

• editing

• 人像生图

• DiT

端到端训练LLM(LMM)#

• Train *

Multimodal Agent*#

Data#

Vision#