Introductional to Traditional Archipelago Foods using The Cpnvolutional Neural Network (CNN)
Abstract
The research with the title Introduction to Traditional Archipelago Foods using the Convolutional Neural Network (CNN) method is an early stage research. In general, this study aims to identify traditional Indonesian foods by comparing 3 CNN models, namely Resnet50,EfficenNetV2M and EfficientNet B6. The method of data collection is by manually collecting images of 20 types of traditional Indonesian food from the internet, each type of food 50 to 80 images, then developing into 20 classes based on the type of food. These images are then used as training models. The technique used is preprocessing or normalizing food image input data by cropping, wrapping. Then resize to a size of 224 x 224 and the image is converted to grayscale for the training process. This study uses a framework to facilitate the creation of deep learning programs, namely Keras Applications as one of the modules in the library that provides various deep learning models and is used to extract features from images. Data analysis was carried out using manual levers to calculate system accuracy in the detection test process. The introduction of traditional archipelago foods is the first stage of research, which will be developed to the composition of food ingredients and finally will count calories. Research on the introduction of traditional archipelago foods to the calculation of calories has not been done in previous studies