Optimizing Food Image Classification Using Black Widow Algorithm and Deep Learning Techniques
Bhavana Jamalpur, Sakthi Priya G, S. Venkat, R. Aarthi, S N Kavitha, G. Charles Babu · 2024
Food image classification is a key application of computer vision and Machine Learning (ML) devoted to mechanically classifying and labelling different food items depending on visual content seized in images. Using sophisticated Convolutional Neural Networks (CNNs) and deep learning (DL) models, this skill permits detection of different dishes, culinary creations, ingredients and raises many practical uses. From supporting nutritional choices and recipe recommendations to menu analysis and food quality control, food image detection updates procedures of food identification, providing valuable insights for users, food experts as well as food industry. Its high potential is to power detection and classification of food items provide benefits in upholding better eating habits, cooking improvement, and enhancing food creation and management procedures. This research proposed Food Image Classification Using Black Widow Algorithm and Deep Learning (FIC-BWODL) model. The developed FIC-BWODL method combines a multifaceted model starting with Contrast Limited Adaptive Histogram Equalization (CLAHE) for image preprocessing, certifying improved image quality and optimum feature extraction. Extraction of features executed by utilizing Capsule Networks (CapsNet) which is famous for their capability to capture difficult classified image patterns. Classification performed via Convolutional Autoencoders (CAE), exactly classifying food images. Moreover, model controls Black Widow Optimization Algorithm (BWO) for hyperparameter tuning that enhances the act and supports it with exclusive features of a dataset. FIC-BWODL methodology provides an important leap in food image detection that provides a strong and exact solution for automated food detection.