Recent advancements in artificial intelligence and computer vision have enabled reliable, non-destructive methods for automated food quality assessment. Fruits are highly perishable, and their freshness significantly impacts consumer safety, market value, and food waste. Traditional evaluation methods, such as manual inspection and sensor-based analysis, are often subjective, time- consuming, and costly. This paper presents an image-based artificial intelligence framework for fruit freshness classification and practical confidence-based expiry days estimation using deep learning. A Convolutional Neural Network (CNN) automatically extracts discriminative visual features, including color variations, texture patterns, bruising, and surface degradation, from fruit images. The system classifies fruits into six categories: Fresh Apple, Rotten Apple, Fresh Banana, Rotten Banana, Fresh Orange and Rotten Orange. A confidence-based rule-driven grading mechanism converts softmax probability scores into actionable freshness percentages, quality grades, and estimated remaining expiry days. The framework is evaluated on a diverse dataset with varying lighting conditions and backgrounds, achieving 95.96% classification accuracy. The proposed system is non-invasive, cost-effective, and suitable for real-time deployment, supporting sustainable food management and reduction of food waste.
Thorat et al. (Thu,) studied this question.