A Real-World Cooking Oil Image Dataset for AI-Based Oil Quality Assessment

Md Mijanur Rahman, S. M. Saleh Ahmed, Md. Abul Bashar, Sumaiya Akter

This dataset contains 8,888 real-world images of cooking oils collected to support Artificial Intelligence (AI) and computer vision research in cooking oil quality assessment. The dataset consists of two cooking oil categories: Soyabean Oil and Mustard Oil. Images were collected from 36 commercially available brands in Bangladesh, including 17 Soyabean Oil brands and 19 Mustard Oil brands. The images were captured under real household cooking conditions using smartphone cameras and manually organized according to oil type, brand, and degradation phase.Soyabean Oil samples are categorized into five degradation phases (Phase 1–Phase 5), while Mustard Oil samples are categorized into four degradation phases (Phase 1–Phase 4). The dataset is intended for image classification, food quality monitoring, machine learning, and deep learning applications. Since the images were collected under real-world conditions, variations in lighting, viewing angles, reflections, and image quality may be present.<b>Categories for this Data:</b><b>Soyabean Oil:</b>Phase 1 (Fresh), Phase 2 (Slightly Reused), Phase 3 (Moderately Degraded), Phase 4 (Highly Degraded), Phase 5 (Severely Degraded)<b>Mustard Oil:</b>Phase 1 (Fresh), Phase 2 (Reused), Phase 3 (Highly Degraded), Phase 4 (Severely Degraded)<br>