Calories are crucial for maintaining good health and preventing diseases including liver disease, cholesterol, cancer, and coronary heart disease. There are more overweight adults in the world compared to underweight persons. The most important risk factor, according to data from India, is abdominal obesity. Thus, keeping track of dietary calories is crucial to living a healthy lifestyle. In today’s fast-growing digital era, Deep learning is one of AI’s subfields that has been in use in food related domain for decades. Deep learning models provide cutting-edge method for analyzing images that guarantees food challenges and finds solutions to issues, since their deeper networks are more adept at processing a wide range of image characteristics. In this paper, we suggest a system that utilizes GAN based segmentation, and Deep learning models InceptionV3, MobileNetV2, DenseNet121 and VGG16 for recognition of food and prediction of food nutrients. This study also focuses on the comparative analysis of the four models with and without GAN based segmentation for concluding on their performance. The performance is calculated on the basis of evaluation metrics comprising of accuracy, precision, recall and F1-score. The models trained with GAN based segmentation approach achieved better performance in comparison to without GAN based segmentation approach and among all these models, InceptionV3 outperformed achieving an accuracy of 99.08% and 98.32% respectively in both these approaches.