CountBench.

CountBench contains 540 images with 2 to 10 instances of a particular object, and captions reflecting the number of objects. It is used for testing and evaluation and has no overlap with the training set C.

Best published scores.

No results indexed yet — be the first to submit a score.

Have a score that beats
this table?

Submit a checkpoint and a reproduction script. We will run it, publish the score, and — if it takes the top — annotate the step on the progress chart with your name.

• 01A public checkpoint or API endpoint
• 02A reproduction script with frozen commit + seed
• 03Declared evaluation environment (Python, deps)
• 04One row per metric declared by this dataset
• 05A contact so we can follow up on discrepancies