<p>Fruit ripeness is typically classified into two to five ordinal categories, but the evaluation criteria vary across studies owing to inconsistent definitions. In this study, a ranking prediction model was developed to classify strawberry ripeness with adjustable thresholds and categories. A web application was designed using a unique rank-based sensory evaluation algorithm that combines the paired comparison and ranking methods. Three panelists ranked 150 strawberry images online, and a prediction model using EfficientNetV2 was trained. A rank-based two-, three-, and four-level classification model was developed by applying the threshold values in ranking prediction. As a result, Kendall’s coefficient of concordance calculated from the sensory evaluation results of the three panelists was 0.86, indicating a strong agreement among the strawberry ripeness rankings. The coefficient of determination (<i>R</i>²) of the rank prediction model was 0.84. Furthermore, the overall accuracy value of the classification model was 0.83 or more for all threshold patterns. As opposed to using fixed categorical labels, our ranking annotation and modeling procedure can be applied to predicting continuous phenomena in fruit quality changes using continuous labeling.</p>