kubricount-subset

Dataset Card for KubriCount (subset) KubriCount is a large-scale synthetic benchmark for multi-grained visual counting, introduced in the paper Count Anything at Any Granularity (Liu, Wu & Xie, SJTU 2026). It reframes open-world counting as a prompt-following problem across five explicit semantic granularity levels, supported by the most comprehensively annotated counting dataset published to date. This is a FiftyOne dataset with 6736 samples. Installation If you… See the full description on the dataset page:

Source: Hugging Face Hub (Voxel51/KubriCount). Metadata imported from the dataset’s Hub tags.