ImageNet Roulette uses a neural network trained on the “people” categories from the ImageNet dataset to classify pictures of people. It’s meant to be a peek into the politics of classifying humans in machine learning systems and the data they’re trained on.
ImageNet Roulette isn't designed to handle heavy traffic so if it's not working for you please be a little patient.
Via Farid Mheir
WHY IT MATTERS: ImageNet has 14M pictures categorized in 32000 categories and is often used as the training dataset for image recognition neural networks. Here this website allows you to upload an image and see how the system would categorize you. It is interesting to play with different images to see what the system returns in order to get a sense of the huge amount of data that AI systems require in order to do a good job. We are far away from systems that can classify any image in any context but if you have a special niche with sufficient amount of data, it is quite impressive what can be done in a relatively short amount of time.