This is the first lecture of Stanford’s CS231n, “Convolutional Neural Networks for Visual Recognition,” in its widely watched 2017 edition introduced by Fei-Fei Li. CS231n is one of the most influential courses in computer vision, and this opening session sets the historical and conceptual stage before the course turns to the technical material.
Li traces the story of vision from its origins in biology through the early decades of computer vision research, explaining how the field repeatedly ran into the limits of hand-designed features. She frames the ImageNet project and the 2012 breakthrough of deep convolutional networks as the turning point that moved the field toward learned representations, a shift she helped drive.
For a reader who wants to understand not just how image recognition works but why it developed the way it did, this lecture provides the narrative that makes the rest of the course coherent. Taught by the researcher behind ImageNet, it carries firsthand authority on one of the pivotal developments in modern AI, and the full course remains freely available.