Count objects found by the neural network

I'm working with neural networks and I need to quantify how many objects there are in an image through the neural network.

For example:

Want to make a classification that can tell me how many balls there are in this image and show which ones are balls.

net = googlenet;

I = imresize(imread('ball.jpg'), [224 224]);

classify(net, I)

Now I want to know how many balls are in this image.

Answer 1

[Antoni Woss]

A good place to start with detecting objects in images using a neural network is through using an object detector network architecture, such as a "You Only Look Once" (YOLO) network. See the documentation example for training such a network to detect vehicles in an image - https://www.mathworks.com/help/vision/ug/train-yolo-v2-network-for-vehicle-detection.html. If you want to count up the number of vehicles the network has detected, you can count the number of rows of the bboxes variable, i.e. size(bboxes,1).