BEEF QUALITY DETECTION APPLICATION USING CONVOLUTIONAL NEURAL NETWORK

Beef consumption in Indonesia increases every year, but this high demand triggers traders to cheat by selling meat that is not fit for consumption, such as rotten meat. This endangers public health and raises concerns, such as the case of the discovery of rotten meat in Makassar. To help people recognise the quality of beef, this research develops a beef quality detection application using the Convolutional Neural Network (CNN) method with Edge Impulse Studio. This application was developed using a prototype model and Research and Development (R&D) method at the Electrical Engineering Laboratory of Makassar State University. Tests showed a model accuracy rate of 99% and test accuracy of 99.17%. Evaluation based on ISO 25010 includes four aspects: functional suitability reached 100% according to the validator lecturer, usability with 97% response proportion, performance efficiency with ±85% CPU usage and 83.3 MB memory, and portability which shows the application runs well on various Android versions. Compatibility was tested by running the application simultaneously with other applications, and the results remained optimal. This application provides a practical solution to detect beef quality quickly and accurately, helping people prevent health risks due to consumption of unfit meat.