Özet:

Abstract Food is important for human survival and has been the focus of several medical conventions. Novel dietary assessment and nutrition investigation technologies now provide more options to assist people in understanding their regular eating habits, researching nutrition trends, and maintaining a balanced diet. People with cancer are now advised to eat a nutritious, balanced diet in order to maintain their quality of life and achieve optimal health results. It is necessary to maintain a healthy body weight and to consume nutritional foods. The major purpose of this work is to provide healthy and nutritious meal suggestions for cancer patients. This research proposes a CapsNet model, a kind of deep neural network, for food recommendation. Preprocessing, feature extraction, and classification are the three key procedures in the proposed study. Reshaping, squash function and routing agreement are the major operations of the proposed work. There are six layers are used to build the CapsNet model which also has the two main layers called Primary Capsule layer and Food Capsule layer. After preprocessing of the food image, the features are extracted. As a final point, depending on the identification outcomes, the system will evaluate the nutritional substances and afford a dietary assessment report by estimating the capacity of fat, calories and carbohydrate. For analysis Food-101 dataset and cancer patient dataset are used. The performance of the proposed methodology is analyzed through different evaluation metrics and achieving an accuracy of 95% and the loss value also decreased. To prove the effectiveness of the proposed methods, this study presents a comparison with the advanced techniques.

Anahtar Kelimeler: