Deep Neural Networks for Image-Based Dietary Assessment

Due to the issues and costs associated with manual dietary assessment approaches, automated solutions are required ease speed up work increase its quality. Today, able record a person's intake in much simpler way, such as by taking an image smartphone camera. In this article, we will focus on image-based approaches assessment. For food recognition problem, deep neural networks have achieved state of art recent years, present our field. particular, first describe method for beverage using network architecture, called NutriNet. This method, like most research done early days learning-based recognition, is limited one output per image, therefore unsuitable images multiple or items. That why that perform segmentation considerably more robust, they identify any number items image. We also two methods - based fully convolutional (FCNs), other residual (ResNet).