Computer vision

• definition :

  • a scientific field that extracts information from digital images

• Computer vision started from the undergraduate summer project in 1966.
  a hard problem : 지금까지도 해결해야 할 부분이 남아있다. 현재진행형...

  • humands do unconsciously, but that is hard for computers

Understanding human vision

• vision comes dow to two components
  1. sensing device
  2. interpreting device
     

Goal of computer vision

• methods to acquire / process / analyze / understand
  • images and high-dimensional data from the real world
  • in order to produce numerical or symbolic information
• bridge the gap between pixel and meaning
  우리가 보는 것은 기차 사진인데, computer 입장에서는 단순한 숫자들의 나열일 뿐이다.

Brief History of Computer Vision

• [1960s] : Picture handling system
  investigate the understanding of visual perception

• [1970s] : Stage of visual representation
  Marr's theory
  

• [1980s] : Recognition via parts
  introduction to object recognition
  generalized cylinder and pictorial structure
  edge
  edge detection
  perceptual grouping(segmentation)

• [1990s] : Recognition via matching
  scale invariant feature transform(SHIFT)

• [2000s] : Face detection
  Harr cascade classifier

• The trends of 2010s :
  high-quality large-scale annotated datasets
  dramatic increase in computational power
  rapid dissemination of ideas
  specialized sensors and hw for computer vision

Four main task of computer vision

1. image classification :
   categorizing the entire image into a class

2. object localization
   해당 object가 image 내에서 어디에 존재하는지

3. object detection
   1, 2를 동시에 수행.
   object가 어디에 있는지 찾고, 어떤 object인지

4. segmentation :

Summary

• computer vision is an interdisciplinary field

• traditional methods aim to acquire high-quality features from pixels
  while modern deep learning automates the feature extraction

• despite our success, computer vision has a long way to go