QUICK SEARCH CELEBRITY BY IMAGE USING DEEP LEARNING

Abstract

The importance of facial recognition systems has increased rapidly over the past few decades. The facial recognition system processes biometric information, but its applicability is easier and the range of operation is greater than that of other processors, ie; scan fingerprint, iris, signature, etc. This study attempts to create a fast celebrity search system using facial recognition. In addition, the system also displays a list of celebrity videos in the database after searching. The above videos were processed before being added to the database. The face-detection and image preprocessing using Multi-task Cascaded Convolutional Network. The feature data is then extracted by FaceNet from the processed images, which are classified by a Learning Similarity algorithm. The list of videos will be stored in a database, which is also used in web applications using Python and Django web frameworks. The place to study the system completely on the Ubuntu operating system. Data is collected from 10 celebrities including videos and photos of their faces. The total number of videos is expected to be 100 videos. The collected information was saved into the system's database while its facial recognition model learned facial images. In addition, users can use the system to quickly search for celebrities by inputting images, which will return videos with celebrities. The tested system has acceptable performance for face recognition and is also capable of detecting and recognizing multiple faces in the resulting images.