3. About Me
Mia Chang
Data Scientist @ Linker Networks
- Microsoft 2017 Data Platform MVP
- Algorithm Research
- Computer Vision
4. Agenda
OpenCV
電腦怎麼看世界
● What’s OpenCV
Computer Vision Workflow
How does OpenCV see things?
● Examples
1. Learn From The Beginning
2. Work with Machine Learning
3. Work with Deep Learning
● Reference, Documents
5. Computer vision is the transformationof data
from a still or video camera into either a decisionor a new representation
6. The input data may include some contextual information
such as “the camera is mounted in a car” or
“laser range finder indicates an object is 1 meter away”.
The decisionmight be “there is a person in this scene” or
“there are 14 tumor cells on this slide”.
A new representationmight mean
turning a color image into a grayscale image or
removing camera motion from an image sequence.
7. See Image / Video File / Stream
Think Context / Rules
Action Decision Message / Signal / New Presentation
Computer Vision is...
Example, Fish Recognition, Image Processing, Streaming Analysis...
8. OpenCV
Provide a simple-to-use computer vision infrastructure
Open source computer vision library
The library is written in C and C++ and
Runs under Linux, Windows, and Mac OS X
Active development on interfaces for Python, Java, MATLAB
9. OpenCV
Features
Written in optimized C++
Take advantage of multicore processors.
Real-time Applications
Application Field
The OpenCV library contains over 500 functions that span many areas in vision,
Including factory product inspection, medical imaging, security,
user interface, camera calibration, stereo vision, and robotics.
OpenCV also contains a full, general-purpose Machine Learning library (ML module).
12. Match the following image formats to their correct number of channels
● RGB
● GrayScale
I. 1 channel
II. 2 channels
III. 3 channels
IV. 4 channels
A) RGB -> I, GrayScale-> III
B) RGB -> IV, GrayScale-> II
C) RGB -> III, GrayScale -> I
D) RGB -> II, GrayScale -> I
13. Match the following image formats to their correct number of channels
● RGB
● GrayScale
I. 1 channel
II. 2 channels
III. 3 channels
IV. 4 channels
A) RGB -> I, GrayScale-> III
B) RGB -> IV, GrayScale-> II
C) RGB -> III, GrayScale -> I
D) RGB -> II, GrayScale -> I
15. What!
Demo 01 - Filter
What did you learn?
1. Play with RGB
2. Play with _B, _G, _R
3. Channels
Learn More:
* Google: IG Filters + Algorithm
* Seven grayscale conversion algorithms
When I use it…
Make more training images
17. What!
Demo 02 - SIFT
What did you learn?
1. SIFT with your picture
2. You can put on a mask (optional)
3. Draw a circle on kp
Learn More:
1. Introduction to SIFT
2. Fun with Frequencies
3. LoG and DoG Filters
4. 圖片的特殊武器: Blob
21. What!
Demo 04 - Tensorflow
What did you learn?
1. How to Analysis Streaming?
2. CV2 cowork with Tensorflow
3. How can I custom one for myself?
Learn More:
* Github: object_detector_app
* Building a Real-Time Object Recognition
App
* Github: Tensorflow object_detection
22. Recap
● What’s OpenCV
Computer Vision Workflow
How does OpenCV see things?
● Examples
1. Filter
2. SIFT
3. KNN
4. Tensorflow
● Documents