Nowadays, automatization of the quality control (QC) process is widely used in industry. In this paper, we propose a convolutional neural network for ordinal classification and quality control of guava fruit. Current state-of-the-art methods primarily address this problem using nominal deep learning classification techniques, which do not take into account the ordinal structure of different classes. Furthermore, they fail to appropriately penalize errors between these classes, which is a crucial consideration for practical applications. To address this problem, our convolutional neural network integrates a probabilistic ordinal link function in the output layer and a loss function that considers the distance between categories, based on the weighted Kappa index. We use a novel dataset from the literature, which includes three different classes of guava fruit, for experimental testing. Results demonstrate that the proposed deep ordinal classification approach improves accuracy and reduces misclassification compared to a standard nominal classification. Specifically, our ordinal model achieves an accuracy of 94% compared to 90% for the nominal version.