Face Detection bad accuracy

I am trying to make a face detection app. I have followed the example of vision.CascadeObjectDetector as done in the example (<http://www.mathworks.com/help/vision/examples/face-detection-and-tracking-using-camshift.html)>. Unfortunatelly the accuracy is not good enough it detects some background or objrcts as faces. How can I improve it ?

Answer 1

[Dima Lisin]

There are several options in the vision.CascadeObjectDetector that you can tweak. If you know how large you expect the faces to be in your images, you can set MinSize and MaxSize to eliminate the false detections that are too small or too big to be a face. You can also try increasing MergeThreshold, or set an ROI (region of interest) to exclude the parts of the image where you do not expect to see any faces at all.

Alternatively, you can also try using a different model: Frontal Face (LBP) instead of the default Frontal Face (CART). You can also try detecting the upper bodies, using the Upper Body model, and then detect the faces inside the resulting bounding boxes.

[Mohammed Hagras]

Thanks for your reply. I will try these silutions.

[Aj_ti]

If I want to detect upper body first, I need to crop the bounding box for upper body then detect face in the cropped upper body image right?

Answer 2

[Justin Pinkney]

You can try this deep learning based face detector: https://github.com/matlab-deep-learning/mtcnn-face-detection

It has much better performance than the built in vision.CascadeObjectDetector

[L H]

How to use it?