DEVELOPMENT OF A HAIRSTYLE APPLICATION BASED ON FACE SHAPE AND SOCIAL MEDIA DATA

• Main Author: Riyanto, Daniel
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/72580
• Institution: Institut Teknologi Bandung

Changing one's hair can often improve one's overall appearance, but there's always the possibility of something going horribly wrong. Unless the new look is tried virtually first. Nowadays, there are apps that help you try out new hairstyles virtually. Most of these applications need to do 3 steps to make the user able to get an image with a new hairstyle. In this Final Project, the author wants to create a hairstyle system/website that can help users virtually try hairstyles by changing the 2nd and 3rd steps from existing applications. The 2nd step that is changed is that the recommendation is replaced with popularity in the community. Then, the 3rd step that is changed is the method that can help users try the recommended hairstyle. In this Final Project, the author wants to try to train face shape classification models from research and other AI models. Later, the best model will be used for step 1 in the hairstyle system created. The suitability of the website results in accordance with the system flow design is tested in three tests, namely Functional Testing, Integration Testing and User Acceptance Testing (UAT). The results of Function Testing show that the API endpoints implemented are in accordance with the endpoint design. Integration Testing results show that the results of website integration are in accordance with the system flow design. The results of UAT conducted on 5 participants showed that 5 out of 5 participants successfully completed 3 out of 3 test tasks and they gave many positive comments so that it can show again that the results of website integration are in accordance with the design of the system flow. The helpfulness of the website was assessed by Intrinsic Motivation Inventory (IMI) subscale value/usefulness assessment and additional questions to 15 participants. The IMI score obtained was 5.3 out of 7, meaning that the participants somewhat agreed that this website was helpful.