Coconut oil has a wide range of applications in the food, cosmetics, and health industries. Ensuring the quality of the product is of paramount importance for the safety of the consumer and the value of the product. Conventionally, the assessment of coconut oil quality has been conducted through manual inspection or laboratory testing. However, these methods can be laborious, costly, and lack precision. In recent years, machine learning has demonstrated considerable potential in the automation of classification tasks. Nevertheless, prevailing models continue to encounter difficulties in terms of precision and the capacity for extrapolation, particularly in relation to the management of real-world variations in data. This research is driven by the necessity for a more expeditious, precise, and economical approach to the classification of coconut oil quality. We propose a deep learning approach, combined with optimization models, to improve classification performance. The primary contribution of this study is the development of a system that utilises deep learning methodologies for the automated analysis of coconut oil images or characteristics, with the objective of evaluating their quality. The standard CNN model achieved 92% accuracy in classifying coconut oil quality. After applying Particle Swarm Optimization (PSO) for hyperparameter tuning, the accuracy improved to 95%. This shows that optimization helps the model learn more effectively, resulting in better performance and more reliable classification results. The following steps are envisaged for subsequent work: firstly, an expansion will be made to the dataset, and secondly, more kinds of oil will be included. In addition to these, the model’s ability to function in different environmental conditions is to be improved. In addition, the objective is to develop a user-friendly application with a focus on assisting small-scale producers in the real-time verification of product quality.