Pixels Activity

Background:

Pixels are the fundamental unit of images and are what machine learning models use to extract information from an image; the two main parts are pixel intensity and pixel resolution. The intensity means how bright the pixel is, while resolution is how many pixels make up the image.

This means that the more pixels our image has the more data our image has and thus the more information our machine learning model can extract from the image. This is why larger resolution images look smoother since there are more pixels to convey the visual information of the image.

However, this also means that it will take more resources and time for a model to look through a higher resolution image compared to a lower resolution image, since it has to go through more information. This is why finding an optimal compromise between pixel data compared to model learning speed is an important aspect of machine learning. If you are using a large data set such as cats for example, you don't need as large of images since the model will be looking more for the general shape of a cat rather than the very fine details, however, if you instead are trying to look at what species of cat then an increase in resolution may be beneficial to extract more fine details to get the species.