Transfer Learning Approach For Predicting The Calorific Value Of Food

Food is the most important part that determines our health, it is vital for us to know what we eat and keep track of our nutrient intake to maintain a healthy lifestyle. In this project, a food recognition system is proposed that detects the food and displays its nutritional contents based on previously stored data. An object detection method is adopted to achieve this purpose. This model is built by training the manually built dataset using TensorFlow API and Single Shot Multibox detection (SSD) MobileNet model pre-trained on COCO dataset. The output is in an easy-to-understand format where it displays the name of the food detected along with its calorific value.