The Open Images Dataset contains about 9 million annotated images with 6,000 category labels, with an average of 8 labels per image. It is divided into a training set of 9,011,219 images, a validation set of 41,620 images, and a test set of 125,436 images. It has more entities than the ImageNet Dataset with 1,000 category labels and can be used for training in the direction of computer vision. This dataset was released by Google, CMU and Cornell University in 2017. The related paper is "OpenImages: A public dataset for large-scale multi-label and multi-class image classification".