Definition
A neural network that applies the transformer architecture to image patches instead of words. Now the backbone of most state-of-the-art vision models.
A Vision Transformer (ViT) splits an image into patches, treats each like a token, and uses attention to relate them. It has largely overtaken older convolutional designs on accuracy at scale and underlies modern detectors, segmenters, and multimodal models.
Also known as
ViT
Specialist software house for video, real-time and AI products. Founded 2005. 50 in-house engineers.
