Train Convolutional Neural Network for Regression with augmented image

Hello

I'am trying to implement a Convolutional Neural Network for Regression with augmented image, and, for this, I try to use this exempla for my dataset:

https://www.mathworks.com/help/deeplearning/ug/train-a-convolutional-neural-network-for-regression.html

But I can't implement data augmentation in this example. What is the simplest way to implement data augmentation for this example?

Answer 1

[Shreeya]

Hi Tiago

To implement data augmentation in the example you've mentioned, you can follow these steps:

1. Load the datase: Load the dataset as per the example
2. Implement Data Augmentation with Image Processing Toolbox: MATLAB provides the Image Processing Toolbox which can be used for various image operations to augment a dataset. In this case, Since you want to predict the angle of rotation of handwritten digits, image operations such as rotation, translation, sheer and scale can help augment the training dataset. The documentation linked below contains the implementation of all these operations:.

https://in.mathworks.com/help/deeplearning/ug/image-augmentation-using-image-processing-toolbox.html

Hope this helps!