الفهرس | Only 14 pages are availabe for public view |
Abstract The Study Aim: building a system that overcomes pose variation in face recognition systems and apply it in educational institutions for recognizing students. The proposed system predicts and recognizes face-pose angles of students during entering the exams. Type of study and methodology: This study combines two types of studies: A. Descriptive Study: This study is based on the methodology of (sample survey (. B. Semi-empirical study: The semi-empirical study is based on applying the proposed system on a sample of students. Study Sample: The dataset of the proposed system consists of (270) images of (30) students in computer teacher preparation department in February 2018 each student has (9) images with different face-poses (-90+0, -60-90, -60+90, +0-90, +0+0, +0+90, +60-90, +90+0, +60+90). Statistical Methods: The researcher employed the statistical treatment appropriate with the nature of the study foremost of which are Accuracy Rate- Error Rate- Mean Absolute Error- Slandered Division. Summary of the study results: The study concludes a host of results, the most important of which are: 1- The MAE for pitch= 5.2 and for yaw = 5.97. 2- The accuracy rate was for SVM (97.53%), but the lowest accuracy rate was for RF (90.12%) in classifying by face-pose. On the other hand, in classifying by face the highest accuracy rate was for SVM (66.67%), and the lowest accuracy rate was for RF (60.6%) 3- There was a statistically significant difference in the mean of face-pose recognition accuracy between the proposed method1, the proposed method2 and also the classic method for the proposed method1. |