Ovis-Aries: A Sheep Breed Image Dataset

Published: 18 December 2023| Version 1 | DOI: 10.17632/cstwjgtxfd.1
Contributors:
,
,
,
,

Description

This dataset encompasses a range of low-resolution images featuring four globally recognized sheep breeds. These breeds were thoughtfully selected from various regions worldwide to ensure a diverse and comprehensive representation. The dataset serves as a visual reservoir, providing a detailed portrayal of the distinctive characteristics of these sheep breeds, thereby aiding in their precise classification. It comprises four distinct categories: Marino, Poll Dorset, Suffolk, and White Suffolk, totaling 1680 original JPG images (with 420 images in each category). The dataset is structured into two variations: one containing original images and another with resized images. The original image dataset has a dimension of 156X181 pixels, while the resized version is set at 224x224 pixels. These images were captured by an installed system at Kingston Rest, a commercial sheep-producing farm in the South-West of Western Australia. The farm visit occurred on May 2nd, 2018, during the autumn season, from 10:00 am to 1:00 pm. The weather conditions were partly cloudy with light intermittent showers, and the average temperature was 15.6 °C. A GoPro Hero5 camera (model: GPCHDHX-502) was utilized to capture images of different sheep breeds. This particular camera model was chosen for its waterproof, rugged design, and its ability to withstand vibrations caused by the drafting machines.

Files

Institutions

University of Western Australia, Bangladesh University of Business and Technology, Universiti Sains Malaysia

Categories

Computer Vision, Image Processing, Machine Learning, Sheep, Image Classification, Agricultural Animal, Deep Learning

Licence