firstoff/animalmind-oxford-pet

AnimalMind — Oxford-IIIT Pet (Normalised)

Baseline dataset for the AnimalMind project.
Derived from pcuenq/oxford-pets (Oxford-IIIT Pet Dataset, Parkhi et al. 2012).

What changed

Original	This version
Single rain split	90/10 rain/ est split (seed=42)
label as raw string	reed normalised (underscores removed, title-case)
dog bool field	species string field: dog or cat
Original image sizes	All images resized to 224 x 224 RGB
Multiple columns	Minimal schema: image, species, reed, split, source

Schema

Column	Type	Description
image	PIL Image 224x224	Pet photograph
species	string	dog or cat
reed	string	English breed name
split	string	rain or est
source	string	oxford_iiit_pet

Statistics

• Total: 7 390 images
• Train: 6 651  |  Test: 739
• Breeds: 37 (12 cat breeds, 25 dog breeds)
• Species (train): ~2 153 cats / ~4 498 dogs

Purpose

This dataset serves as the Phase 1 validation baseline for AnimalMind.
The goal is to validate the full ML pipeline (data → model → API) before moving to the proprietary AnimalMind dataset focused on behaviour and emotion.

Citation

@InProceedings{parkhi12a, author = {Parkhi, O. M. and Vedaldi, A. and Zisserman, A. and Jawahar, C.~V.}, title = {Cats and Dogs}, booktitle = {IEEE Conference on Computer Vision and Pattern Recognition}, year = {2012}, }