Optimizing CNN for Cuisine Image Recognition with Adam Optimizer and Dropout

Cuisine represents the unique culinary traditions and cultural expressions of different regions, playing a significant role in both cultural heritage preservation and the burgeoning field of culinary tourism. This study focuses on optimizing Convolutional Neural Network (CNN) performance for Padang cuisine image recognition using a dataset from Kaggle, comprising nine distinct food categories. We enhance CNN performance by employing the Adam optimizer and dropout techniques to address challenges in recognizing complex dishes with varying ingredients, textures, and colors. Through systematic experimentation, the results demonstrate that a 0.2 dropout rate combined with the Adam optimizer achieves the highest accuracy of 87% and a validation accuracy of 76%, surpassing other optimizers such as Adagrad and SGD. These results demonstrate the effectiveness of the Adam optimizer in improving CNN accuracy for cuisine image recognition. This research provides valuable insights for optimizing CNN architectures in food recognition technology, specifically for Padang cuisine.