ImageNet

ImageNet is a large visual database designed for use in visual object recognition research. Key facts:

• Created by researchers at Stanford University and Princeton University in 2009.
• Contains over 14 million images across 20,000 categories.
• Categories include animals, vehicles, furniture, clothing, and more.
• Has roughly 1000 images per category with labels.
• Images are high resolution and diverse.
• Used to train and benchmark deep learning models.
• Enabled breakthrough results in computer vision.
• Underlies models like AlexNet and ResNet.
• Raises ethical issues around depicting and labeling people.

ImageNet played a pivotal role in the rise of deep learning due to its scale and image diversity. The annual ImageNet competition motivated progress on models like convolutional neural networks that could achieve human-level accuracy.

However, ImageNet has faced criticism for perpetuating offensive stereotypes through some category labels and lack of consent. This highlights the need for caution in data curation.

See also:

• Computer vision
• Convolutional neural network
• Object recognition
• Image classification