Google Unveils Massive Open Images Dataset V4 for Computer Vision
Google has unveiled the fourth iteration of its Open Images Dataset. This comprehensive resource, designed to aid computer vision research, boasts an extensive collection of annotated images.
The dataset comprises over nine million images, meticulously labeled for training purposes. It categorises objects into 600 distinct classes, providing a rich and varied source for model development.
To ensure the robustness of models, the dataset includes a validation set of over 40,000 images. Moreover, it features a test set of more than 125,000 images, each verified by human annotators to maintain accuracy.
The fourth version of Google's Open Images Dataset, with its vast and diverse collection of annotated images, stands as the largest publicly accessible dataset of its kind. It is poised to significantly contribute to advancements in computer vision research.
