Crystal Guava Grading 1000 (CGG-1000)

Résumé fourni par la source

This dataset consists of 1000 high-resolution JPEG images of crystal guava (Psidium guajava) fruits, collected to support research in automated fruit grading and machine vision applications in agriculture. Each fruit was photographed from two fixed angles (top view and side view) using synchronized Logitech C920 Full HD webcams under controlled lighting conditions within an LED-illuminated imaging box. A total of 500 guava fruits are included, each represented by two images (top and side). Every fruit is labeled with its measured weight (in grams) and an expert-assigned quality grade: - Grade A (premium quality) - Grade B (standard quality) - Grade C (small size or minor defects) - Reject (unfit for market) The ground-truth labels were determined by trained agricultural experts following commercial grading criteria based on size, shape, and external appearance. Fruit weights were recorded using a calibrated digital scale. Four metadata files (CSV format) — one for each grade — accompany the images, listing the filename, grade, camera position (top/side), and weight for each image. This dataset is especially suitable for training and benchmarking deep learning models for classification and regression (e.g., weight prediction).

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.