Bell pepper shelf-life assessment relies on subjective visual inspection, yet quality indicators respond differently under contrasting storage conditions, limiting single-indicator approaches. This study introduces an integrated framework for objective shelf-life determination using YOLO11x-based damage detection and weight loss measurement. A pretrained YOLO11x model was fine-tuned on 996 images of 40 red bell peppers stored at 8℃ for 25 days. Through iterative refinement and hard example mining, the model achieved a damage mean Average Precision (mAP) of 0.726 at an IoU threshold of 0.5. For validation, red bell peppers of a different cultivar, harvest year, and growing site were stored at 8℃, and at 4℃ followed by transfer to room temperature (4℃ to RT), for 20 days each to simulate chilling injury, which substantially accelerates surface damage. A Quality Index (QI) was established by integrating the YOLO-derived damage ratio (threshold of 5.4 percent) and weight loss (threshold of 7.0 percent), each given equal weighting of 0.5. The QI showed a high coefficient of determination with visual quality scores (R² = 0.9861), identifying the commercial acceptability limit at a QI of 1.0 or higher. Furthermore, when transfer learning was applied to yellow bell peppers, a second cultivar, the model retained 92.0% of its performance using only 12.7% of the training data, with the QI maintaining strong correlation with visual quality scores (R² = 0.9852). This approach balances the sensitivity of visual assessment with the stability of weight measurement, providing an objective, reproducible framework valid across contrasting storage conditions, cultivars, and harvest seasons.