Deep Learning 활용 사례

Computer Vision Tasks (1)

  • No object Just pixels
    • Semantic Segmentation
      • CAT | GRASS | TREE ..
  • Single object
    • Classfication
      • CAT
    • Classfication + localization
      • CAT
  • Multiple objects
    • Object detection : Videos ..
      • CAT | DOG | DUCK
    • Instance segmentation
      • CAT1, CAT2, DOG, DUCK

Computer Vision Tasks (2) Obtical Character Recognition

  • 주민등록증
    • 이름
    • 주민번호
    • 주소
    • 발급일자
    • 발급기관
  • 상품
    • 상품명
    • 회사
    • etc.

Computer Vision Tasks (3) Motion Capture without Marker

  • Object detection
  • etc.

Natural Language Processing Tasks

  • Language generation
  • Answering questions
  • Text classfication
  • Sentiment analysis
  • Machine translation
  • 예: ChatGPT

Multi-Modal Tasks (1) Image Generation (text to image)

  • Multi-Modal : Text + Vision + Audio + etc.
    • script : "A person half Yoda half Gandalf"
      • process : Embedding -> Embedding + DM - Denoise + ... + -> result(image)

Multi-Modal Tasks (2) Video Generation (text to video)

  • 예 : Sora AI

출처 : 메타코드M

+ Recent posts