Background The most common dental diseases are tooth decay (caries) and gum disease (gingivitis and periodontitis). Obviously, these diseases are caused by dental plaque (bacterial biofilm). Although most patients brush their teeth every day, they cannot keep all their teeth clean. Areas in the mouth that are difficult to access, such as crowded areas, posterior teeth or interdental areas, are usually affected (site-specific). After a thorough professional tooth cleaning, dental plaque will begin to accumulate on the tooth surface near the gum edge within a few days. Clinical studies indicating that regular disruption to the plaque is needed and can prevent and arrest gum disease. However, dental diseases may take years to develop, the patient usually does not have any pain symptoms unless the disease has progressed to the advanced stage. A significant amount of resources and clinical time have been used to motivate and instruct patients to keep their mouth clean and yet the results are not satisfactory. It is desirable to adopt an automated technique for monitoring oral health daily so participants can seek treatment when it is needed. Patients' response to plaque accumulated at the gum margin is by inflammation which brings more blood cells to the site to fight against the bacterial invasion. Inflammation of gum is manifested as an increase in redness (color), an increase in volume (oedema), and loss of surface characteristics (stippling; gum fibre attachment). These affected areas can be identified by visual inspection with the dentist during the consultation or using intraoral photography. The objective of this research is to apply deep neural network technology to detect gum inflammation from intraoral photos. As the target inflammation site is at gum margin with varied shape and size, semantic segmentation at pixel level is needed. In this research, the investigators are planning to have an extensive study of deep neural network (DNN) approach for the automatic multiple level gum disease detection. Standardized intraoral photography will be collected for 1200 cases and will be labelled by several dentists as diseased (inflammation), healthy or questionable. Only gum area in which the dentists have same rating will be used to train/validate the system. Using the successfully developed system, one can use his/her mobile device to monitor their gum health when needed. They may be able to prevent the two main oral diseases (tooth decay and gum diseases) with minimal additional cost. It will be an important contribution to the promotion of public dental care. Aim of study This study aims to train and validate the computer to automatically monitor gum inflammation using standardized intraoral photos and selfie by smartphone. 1. to collect 1200 standard intraoral photographs and randomly cropped into training and validation sets. 2. to develop ground truth gingivitis label images into four health status levels (healthy, questionable healthy, questionable diseased and diseased) and verified by dental specialists. 3. to develop intelligent system for automatically detect inflamed disease sites with four health status levels. 4. to develop and standardize the image acquisition protocol for the detection with mobile devices. Hypothesis A diagnostic tool should be able to diagnose true disease and true health which described as sensitivity (positive when true disease) and specificity (negative when true health). The primary outcome will be the area under the receiver operating characteristic (ROC) curve (AUC). The hypothesis of this study is the trained gingival detection system is able to detect the changes of gum inflammation with high sensitivity and specificity.