Food Item Recognition, Calorie Count and Recommendation using Deep Learning

  • Apurv Upreti, Dr. C. Malathy

Abstract

Due to recent advancements in the field of deep learning, image recognition has become a very active field of research. Lots of applications have started using deep learning to solve complex problems. In this paper our focus would be to use deep learning techniques for recognizing food images from FOOD101 dataset. Food items vary a lot from place to place, it is thus not possible to collect information about all food items even for a small region. So we will focus on only 101 food items. We would use transfer learning in the inceptionv3 model. Inceptionv3 model is a deep learning model developed by google with excessive training over images. After preparing our model we will check for the calorie count of the food item using a pre-stored dictionary. We also train a simple linear model for calculating the BMI of a person. Then using these details we are able to calculate which food item is good to consume and which items must be avoided. This would enable people to keep track of their calorie intake and avoid heavy calorie food items.

Keywords: Food recognition, Calorie count, recommendation, CNN, inceptionV3, Indian Food analysis, BMI

Published
2020-05-05
How to Cite
Apurv Upreti, Dr. C. Malathy. (2020). Food Item Recognition, Calorie Count and Recommendation using Deep Learning. International Journal of Advanced Science and Technology, 29(06), 2216 - 2222. Retrieved from https://sersc.org/journals/index.php/IJAST/article/view/13510