I am processing spam comments by processing the image of the comment data. I want to augment this image data

The reason why spam is processed by processing the image of the comments is because,

1. spam comments use an emoji.
2. Although it does not have meaning in text data, it writes comments with meaning when viewed by humans.Considering these two things, we are classifying images as cnn to spam comments.So my question is how do I augment the data in the context of my project?

I am a high school student and live in a non-English speaking area, so I would appreciate it if you could answer in an easy sentence.

I don't think simply increasing or smashing the image is in line with the situation of my project.