Autonomous System For Attendance And Report Generation Via Email Using Facial Recognition

Abstract

Attendance system plays a significant role in all academic organizations to verify student attendance rates. Attendance is manually taken in the educational environment by calling their student ID or names and being registered as evidence in attendance registers provided by the department heads. This approach is repetitive, complicated work and leads to errors because few students can check attendance for their friends, and lecturers cannot double-check these situations. In comparison, this strategy makes it more challenging to track all students and individuals entering a big classroom setting. The main idea of this thesis is to research and incorporate the current facial recognition methods on the small dataset to check student attendance on an autonomous system. The thesis has total of three main tasks to complete the research and implementing the autonomous system based on the face recognition method on the small dataset. Data acquisition, which is the first task in the thesis, takes an extreme effort to complete. The collection process requires capturing the International University facial images, then minimizing and reshaping the Labelled Faces in the Wilds dataset in order to fit the size of IUStudent dataset. The next step is training and testing models then comparing these models based on the face recognition algorithms with models based on the face verification process, which is the most crucial part of this thesis. The proposed models are used to implements the autonomous system for attendance and Report genration via Email. My proposed models achive a significant performance on the small dataset such as the LFW dataset and IUStudent dataset, at 94.99% and 91.46% respectively. This has a significant improvement compared to traditional face recognition algorithms about 10% due to LBHF algorithm and 36% due to PCA and SVM. Furthermore, The scope of this thesis is relatively minimal while the scope of the face recognition applications is extremely wide. Therefore, I will keep researching and implement the AI-based camera to increase the attendance rate in the academic environment.