International Journal of Scientific & Engineering Research Volume 11, Issue 4, April-2020 1770 ISSN 2229-5518 IJSER © 2020 http://www.ijser.org SURVEY ON APPLICATIONS FOR PREDICTING CALORIES AND NUTRITIONAL VALUES FROM FOOD Rachoti Biradar 1 , Rahul Thiru 2 , Ravi Purwar 3 Guided By : Latha N R 4 Abstract - Recently, people are becoming used to a modern lifestyle since they can be fully consumed by busy schedules at work and at home. Obesity in adults is becoming a common problem. The main cause of obesity is a combination of excessive food consumption and lack of physical activities. Consuming food with a high amount of calories can cause several problems for our health. There is widespread nutritional information and guidelines that are available to users at their fingertips on the internet. However, such information has not prevented diet related illnesses or helped patients to eat healthily. In most cases, people find it difficult to examine all of the information about nutrition and dietary choices. Furthermore, people are obvious about measuring or controlling their daily calorie intake due to the lack of nutritional knowledge, irregular eating patterns or lack of self-control. Recording the amount of calorie intake during each meal is a tedious task. Although people can record their meals and discuss with doctors or experts, it is not very convenient for them to know the number of calories before the meal. Our goal is to empower users by a convenient, intelligent and an accurate system that helps them become aware of their calorie intake and also find the individual nutrients content in the food item. Keywords - Machine Learning, Data Analytics, Neural Networks, Linear Regression, Multiple Linear Regression, Data Mining, Decision Trees, Food. ——————————  —————————— 1. INTRODUCTION Food image recognition is one of the crucial applications used these days. It allows smartphone users to know the name of the food. Many people are interested in tracking what they eat to help them achieve weight loss goals or manage their diabetes or food allergies. This is quite important for travellers who travel to foreign countries. However, most current mobile apps (MyFitnessPal, LoseIt, etc) require manual data entry, which is tedious and time- consuming. Consequently, most users do not use such apps for very long. Furthermore, amateur self-reports of calorie intake typically have an error rate that exceeds 400 calories per day. Several previous approaches rely on an expert nutritionist to analyze the image offline (at the end of each day). Other approaches use crowdsourcing to interpret the image, instead of an expert. However, crowdsourcing is both costly and slow, which hinders widespread adoption. We mainly focus on Indian food as no app in the market currently caters to it. Several existing works do use computer vision algorithms to reason about meals but only work in laboratory conditions where the food items are well separated and the recognition, however, has been 1. Rachoti Biradar is currently pursuing bachelor’s degree in Computer Science & engg in BMS College of Engg, Bangalore. Email : 1bm16cs075@bmsce.ac.in 2. Rahul Thiru is currently pursuing bachelor’s degree in Computer Science & engg in BMS College of Engg, Bangalore. Email:1bm16cs077@bmsce.ac.in 3. Ravi Purwar is currently pursuing bachelor’s degree in Computer Science & engg in BMS College of Engg, Bangalore. Email : 1bm16cs080@bmsce.ac.in 4. Latha N R is Assistant Professor at Computer Science department BMS College of Engg, Bangalore. Email : latha.cse@bmsce.ac.in mainly focused on the correctness of the food name for the given food image. Many techniques are applied, for example using image segmentation to separate the food from the background image. This technique will increase the effectiveness of food identification. In this paper, we take some initial steps towards such a system. Our approach utilizes several deep learning algorithms, tailored to run on a conventional mobile phone, trained to recognize food items and predict the nutritional contents meals from images taken “in the wild”. 2. LITERATURE SURVEY A. NU-InNet: Thai Food Image Recognition Using Convolutional Neural Networks on Smartphone Since AlexNet[2] and GoogLeNet[3] the storage space required for the model is huge these people proposed their model which is NU-InNet[1] (Naresuan University Inception Network), which is helpful for mobile applications because, in mobile applications, Processing time and storage space should be minimum.so the inception module adopted by GoogLeNet was adopted and further improved by keeping the accuracy at the same level. In that NU-InNet model, they have created two versions I.e., NU-InNet 1.0 and NU- InNet 1.1 so in NUInNet 1.0 model the GoogLeNet inception model is modified by changing its 3×3 max- pooling layer and 1×1 convolutional layer to be 1×1 and 7×7 convolutional layers, respectively, and in NU-InNet 1.1 network the NU-InNet 1.0 is modified by changing [4,5] any 5×5 convolutional layer to be 2 3×3 convolutional layers and changing any 7×7 convolutional layer to be 3 3×3 convolutional layers By doing this the accuracies of the models increased compared to the models which own at ILSVRC, which is helpful for the mobile applications. B. A New Deep Learning-based Food Recognition System for Dietary Assessment on An Edge Computing Service IJSER