What is Computer Vision? (2)

#python #pytorch #computervision #deeplearning

*My post explains Image Classification(Recognition), Object Localization, Object Detection and Image Segmentation.

Computer Vision is the technology which enables a computer to understand and analyze the visual things such as images, videos, etc.

(1) Keypoint Detection(Landmark Detection):

can localize the distinctive points of the objects or interest regions in an image: *Memos:
- The image can be one frame in a video.
- Object Localization or Image Classification and Object Localization(Object Detection) is used.
can is used to recognize face, its emotion, the pose of person, clothes, etc.

(2) Image Matching:

can determine that the objects and interest regions in multiple images are the same or similar by comparing their distinctive points.
is the advanced technology of Keypoint Detection.
can be used to match faces, finger prints, etc.

(3) Object Tracking:

can track the objects and interest regions in a video(multiple frames captured in sequence) with bounding boxes. *The video can be multiple images captured in sequence.
is the advanced technology of Object Detection.
is used for surveillance, autonomous vehicles, virtual reality, manufacturing, unmanned stores, etc.

(4) Optical Flow:

can detect and visualize the motion(velocity) of the objects and interest regions in a video(multiple frames captured in sequence): *Memos:
- Velocity is the spped and direction of a moving object.
- The video can be multiple images captured in sequence.
is the branched technology of Object Tracking.
is used for surveillance, traffic monitoring, motion analysis, etc.

(5) Stereo Matching:

can create a 3D scene from two or more 2D stereo images by computing disparity: *Memos:
- A 2D stereo image is the similar image taken from a different position.
- Disparity is the distance between two or more corresponding pixels in the stereo images.
is used for robotics.