<![CDATA[Conventional assessment of fish freshness remains subjective, resulting in inconsistent evaluations. Advances in computer vision and Convolutional Neural Networks (CNNs) offer an opportunity to develop automated and objective approaches for assessing fish freshness through mobile devices. This study developed Fish Check, a smartphone-based application that utilizes the MobileNetV2 architecture to classify the freshness status of marine fish from digital images captured using a smartphone camera. A dataset comprising 2,400 marine fish images was collected. Image acquisition was conducted under varying lighting conditions, camera angles, and object positions to represent practical field conditions. The proposed approach involved image preprocessing, data augmentation, transfer learning-based model training, and performance evaluation using Accuracy, Precision, Recall, F1-score, and a Confusion Matrix. The optimized MobileNetV2 model was subsequently integrated into a React Native mobile application using TensorFlow Lite to enable efficient real-time inference directly on smartphones. The experimental results showed that the proposed model achieved an Accuracy of 96.39%, Precision of 96.11%, Recall of 96.74%, and F1-score of 96.42% demonstrate that the model can effectively distinguish between fresh and non-fresh marine fish. Overall, Fish Check provides a practical approach for real-time fish freshness assessment by combining a locally collected marine fish image dataset with an efficient deep learning architecture optimized for mobile deployment.]]>
Copyrights © 2026
