KubriCount is a large-scale synthetic benchmark for multi-grained visual counting, built for the research project Count Anything at Any Granularity. The dataset targets open-world counting settings where the intended counting granularity must be explicit. It was created by author liuchang666 and was last updated on the Hugging Face platform in May 2026.
Use Cases
- Benchmarking visual counting models based on the dataset's multi-grained query structure.
- Training models to count specific identities or attributes based on the explicit granularity targets.
- Evaluating model robustness against controlled distractors as described in the dataset's purpose.
- Researching instance-level counting for categories or broader concepts as outlined in the project.
Strengths
- Dataset is described as 'large-scale', suggesting substantial size for model training and evaluation.
- Provides controlled distractors and dense instance-level annotations, as mentioned in the description.
- Specifically designed for open-world counting with explicit granularity, offering a targeted research focus.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count, file formats, and license are unknown, which may limit suitability assessment.
- Data is synthetic, which may limit direct applicability to real-world visual scenes.
Provenance
- Source
- Hugging Face, uploaded by author liuchang666.
- Collection Method
- Synthetically generated for the 'Count Anything at Any Granularity' research project.
- Time Range
- null
- Freshness
- Last updated 2026-05-13 03:03:26; freshness should be verified.
- Geography
- null