![](https://crypto4nerd.com/wp-content/uploads/2023/06/1DIFEA2d7iEVSGGf56Fuh6Q-1024x655.png)
The application I want to make requires computer vision technology, so I searched for it and
found out about the Vision framework, so I briefly summarized it.
The Vision Framework is a framework that was announced along with coreML at WWDC in 2017.
- face detection
- Ability to track facial landmarks
- Like iPhone’s LiveText, the ability to find text in a photo
- General image registration, tracking function
Computer vision algorithms can be applied to an image or video to perform various functions such as the above.
openCV is also one of the very popular computer vision frameworks.
- It supports several languages, but unfortunately swift is not supported.
However, since the vision framework was developed by Apple, it supports swift.
Also, it’s my guess, but since it was developed by Apple, isn’t it more compatible with swift? 🤔 - OpenCV is an external framework, so it needs to be installed separately, but the Vision framework is
import Vision
finished with just one code.
Pipelines are made up of subparts
request
handler
VNImageRequestHandler
: Used when working with still imagesVNSequenceRequestHandler
: Note the sequence. Used when working on image sequences (videos)
result
VNRequest
A request made with a subclass of is passed to the handler.- handler can be of
VNImageRequestHandler
typeVNSequenceRequestHandler
- The processed result
VNObservation
is returned as a subclass of .
For example, if a type VNRequest
that is a subclass of is received as a request, a type that is a subclass of is returned as a return.VNDetectFaceRectanglesRequestVNObservationVNFaceObservation
For another example, it is a format VNDetectContoursRequest
where is received as a request and VNContoursObservation
is returned as a return.