Semantic Food Segmentation Using Convolutional Deconvolutional Pyramid Network for Health Monitoring

—This paper presents semantic food segmentation to detect individual food items in an image in the context of Food Recognition (FoodRec) project. FoodRec aims to study and develop an automatic framework to track and monitor the dietary habits of people, during their smoke quitting protocol. Studies have shown a strong correlation between dietary habits’ changes of individuals and smoking cessation process. Abstinence from smoking is associated with several negative effects such as gain of weight, eating disorders, mood changes, and irritability during the initial period of smoke quitting. In this contribution, a novel Convolutional Deconvolutional Pyramid Network (CDPN) is proposed for food segmentation to understand the semantic information of an image at a pixel level. This network employs convolution and deconvolution layers to build a feature pyramid and achieves high-level semantic feature map representation. As a consequence, the novel semantic segmentation network generates a dense and precise segmentation map of the input food image. Furthermore, the proposed method achieved competitive results with 91.77% mean Intersection over Union (IOU) on TrayDataset and 77% mean IOU on MyFood dataset when compared to the state-of-the-art techniques.