AIExplainer

What is stride?

The number of pixels or units that a filter moves over an image or data in a single step

In artificial intelligence, particularly in convolutional neural networks, stride refers to how many pixels or units a filter moves when scanning an image or data. A stride of 1 means the filter moves one pixel at a time, while a stride of 2 means it moves two pixels at a time, and so on.

Imagine a window that moves over a piece of paper with text on it. If the window moves one word at a time, that's like a stride of 1. If it moves two words at a time, that's like a stride of 2. The stride determines how much of the text the window 'sees' at each step.

Self-driving cars use convolutional neural networks to process images from cameras. The stride used in these networks can affect how well the car detects objects, such as pedestrians or other cars, at different distances.

Stride is used in convolutional neural networks to control the resolution of the output. A larger stride reduces the resolution, which can help reduce the computational cost, while a smaller stride increases the resolution, which can help improve the accuracy.

One common misconception is that a larger stride always leads to better performance. However, a larger stride can also lead to loss of important details, which can negatively impact the accuracy of the model.

The concept of stride originated in the field of image processing, where it was used to describe the movement of filters over images. With the development of convolutional neural networks, stride became a key hyperparameter that can significantly impact the performance of the model.

step size filter step

Three products for different needs — explore what’s relevant to you.