File Exchange

image thumbnail

Designing Object Detectors for Real Case

version 1.0.1 (17.9 MB) by Kevin Chng
Method 1 - Image Processing - Colour Thresholding Method 2 - ACF Method 3 - Faster R-CNN

15 Downloads

Updated 16 May 2019

View License

One of the important field in Artificial Intelligence is object detection. There are many approaches in MATLAB. In my view, they are classified into three broad categories.
(1) Image processing/ComputerVision - Color Thresholding, Blob Analysis, Histogram of Gradients, Speeded-Up Robust Features.
(2) Machine Learning - Cascade Object Detector (Viola-Jones Algorithm), Aggregate Channel Features (ACF)
(3) Deep Learning - YOLO v2, R-CNN, Fast R-CNN and Faster R-CNN

In this example, it demonstrates one method from each categories to solve a real-world problem.
1) Method 1 : Image Processing - Colour Thresholding
- Learn basic image processing technique : Extract colour, Difference between Color Space, Morphologically -Open Image, Dillate Image, Calculate Object in Binary Image
- Image Processing App in MATLAB - Color Thresholder
- Limitation of this application

2) Method 2 : Aggregate Channel Features (ACF)
-Learn how to label image using Image labeler App (GUI)
-Train ACF object detector
-How to fine tune ACF accuracy (Remove low scores detection & Overlap detection)

3) Method 3 : Faster R-CNN
-Learn how to label image using Image labeler App (GUI)
-Train Faster R-CNN object detector
-How to fine tune ACF accuracy (Remove low scores detection)

Cite As

Kevin Chng (2019). Designing Object Detectors for Real Case (https://www.mathworks.com/matlabcentral/fileexchange/71522-designing-object-detectors-for-real-case), MATLAB Central File Exchange. Retrieved .

Comments and Ratings (3)

Updates

1.0.1

Add one more method : Faster R-CNN

MATLAB Release Compatibility
Created with R2019a
Compatible with any release
Platform Compatibility
Windows macOS Linux