Deep Learning based Performance Analysis of Nutritional Contents in Food Images

In the present day scenario, analysis of nutrition and calories in daily food intake has become
indispensable. The increasing obesity problems have made a significant effect on the people to be
concerned about the calories they consume. In this paper, we propose a food calorie measurement
system that can help them to measure and manage daily food intake. This method is employed to
identify if the food image is good or rotten. If it is found to be good then it is taken for calorie
measurement analysis and classified based on standard calorific tables using Self-Adaptive Resource
Allocation Network [SARAN]. Then, based on the BMI of a person, the result alarms about whether
the food under analysis is suitable to the person or not. Improvement in the performance analysis
were carried out on ALEXNET Architecture based On Deep Convolutional Neural network. The results
show that the accuracy of the system is acceptable and it will greatly improve and facilitate current
manual calorie measurement techniques.