Abstract:
A haircut can significantly affect anyone’s appearance along with other facial features. It is important to choose a good haircut for women because it can enhance their beauty, personality and confidence. The face shape of the individual is one of the most crucial aspects to take into account while selecting the best haircut. However, in 2020 and beyond, the opportunities to wear a face mask at public places have increased all over the world. Aligned with it, the difficulty to identify the face shape while wearing a face mask at the hair salon has increased. A face mask hides the mouth, nose, chin, and facial muscles, making it difficult for hairdressers to recommend suitable haircuts by identifying the face shape of their client while wearing a face mask at the hair salon. In this research study, introduced such a computer-aided system which was suitable and beneficial to the beauty industry during this pandemic. This novel approach used both Convolution Neural Network and Resnet Architecture to build a face reconstruction model and using transfer learning built a face shape classification model. When uploading a masked face image, the system model reconstructs the hidden parts of the face and shifts the constructed face image to classification model. Additionally, Age, Hair Type and Hair Length are taken as user input. According to those, the system will provide suitable haircuts. The accuracy of the face reconstruct model is 94% and the accuracy of the face shape classification model is 96%.
