Locally Nuisanced Data: ImageNet, Cars, and CUB for Open Set Recognition Testing

A dataset for evaluating nuisance novelty to expose gaps in open set recognition reliability. It is based on established computer vision datasets including ImageNet, Cars, and CUB. The dataset appears to be designed for testing the robustness of image classification models.

Use Cases

• Test model robustness against novel nuisance factors based on the described evaluation of nuisance novelty.
• Benchmark open set recognition reliability based on the dataset's stated purpose.
• Identify gaps in image classification models based on the exposure of reliability gaps mentioned in the description.

Strengths

• Built upon established benchmark datasets (ImageNet, Cars, CUB), which suggests a foundation of known quality.
• Focuses on a specific research problem (nuisance novelty in open set recognition), providing a targeted testbed.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and size are unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

Kaggle

Collection Method

Derived from ImageNet, Cars, and CUB datasets for a specific research evaluation.

Time Range

null

Freshness

Last updated date is unknown; freshness unverified.

Geography

null