How to create image dataset for trainig the neural network?

Hello everyone, ı have to create a image dataset .mat file for image recognition. I have many questions and I am searching for days.

So basically ı created 2 file which includes 13 jpg images for each. Then ı converted them to .mat file. But what ı want is that ı want converting these two .mat file into single one one of them is for training and other one is for test data. However ı couldnt manage them. Then I continued to search and I realized that ı have to use labeling for recognition? then preprocessing and now ı am so confused. Before this project ı worked with mnistdata set but it is like a library ı mean it is already created data set. I needed a solution or referance to work with. Anybody for help please? I hope ı explained my problem.

Thanks in advance

Answer 1

[Amit Dhakite]

Hi Aybuke,

I understand that you are unable to prepare dataset for image recognition and divide them into training and testing data.

In order to do so, you can consider the following steps:

1. Prepare dataset: You can utilize the Image Labeler App to create single .mat file.
2. Preparing training and testing dataset: You can use the .mat file from above step to divide it into training and testing data, by simple matrix slicing.

For your reference, I am providing an article that will help you to train image recognition model: https://www.mathworks.com/help/vision/ug/object-detection-using-yolo-v3-deep-learning.html

To know more about Image Labeler App and matrix slicing, kindly refer to the following links:

1. Image Labeler App: https://www.mathworks.com/help/vision/ref/imagelabeler-app.html
2. Matrix indexing and slicing: https://www.mathworks.com/company/newsletters/articles/matrix-indexing-in-matlab.html