Skip to content
BENIBIX Learn

ImageNet

AI🧭 FoundationsAI History

Also known as ImageNet Dataset, ImageNet Challenge, ILSVRC

A huge set of labelled photos. A contest on it set off the deep learning boom.

Think of it like

It is like a huge picture book. Each photo has its name under it.

Example

A team trains its model on ImageNet photos of dogs, cars and cups. Then it tests how often the model names new photos right.

Why it matters

It showed that big data plus neural networks could beat hand-built methods. Most of modern computer vision grew from there.

Definition

ImageNet is a huge set of photos, each labelled with what it shows. Researchers built it so computers could learn to see. From 2010, a yearly contest asked teams to name the objects in its photos. In the 2012 paper that made deep learning famous, a neural network trained on 1.3 million ImageNet photos in 1,000 classes beat earlier results by a wide margin. That result started the boom in deep learning.
In 2010 the ImageNet contest began. Deep learning later won it by a mile.

Common mistake

Thinking ImageNet is a model. It is the data the models learn from and are tested on.

Key takeaways

  1. 1
    Labelled photos. Each one says what it shows.
  2. 2
    A yearly contest. Teams raced to name objects.
  3. 3
    Sparked deep learning. Big data met neural nets.

Learn it properly