CTHM-AI: Thermal Images and Tabular Data for 74 Cats in Clinical Settings

74 unique feline subjects captured in controlled indoor clinical settings to eliminate environmental thermal noise. The dataset, created by Mohammad Abdulghafar and updated in May 2026, is a standardized derivative of the Thermal Imaging Cats’ Dataset. It provides 7,588 images paired with synchronized 11-feature tabular CSVs, split across 50 training, 12 validation, and 12 testing subjects.

Use Cases

• Train binary classifiers ('Healthy' vs. 'Sick') for cats based on thermal imaging data.
• Develop multimodal fusion models using paired image directories and 11-feature tabular CSVs.
• Benchmark patient-level data splits to prevent data leakage in medical ML studies.
• Evaluate thermal-safe augmentation techniques that preserve raw diagnostic temperature data.

Strengths

• Strict patient-level split (50/12/12) prevents data leakage across subjects.
• Thermal-safe deterministic augmentation yields 7,588 training images to address class imbalance.
• Multimodal design pairs images with synchronized 11-feature tabular CSVs for evaluation parity.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the tabular data is unknown, which may limit suitability assessment.
• Data may reflect bias inherent to the specific clinical settings and source platform.

Provenance

Source

Derived from the Thermal Imaging Cats’ Dataset (TICD).

Collection Method

Rigorously standardized and augmented from the original source, captured in controlled indoor clinical settings.

Time Range

null

Freshness

Last updated 2026-05-07 18:06:19; freshness should be verified.

Geography

null