Hello I am pursuing my B.Tech in computer science engineering. I have made a project on frontal face recognition. I have used Haar Cascade machine learning algorithm for face detection and Linear Binary Pattern Histogram for face recognition. However the system is not robust. Meaning it does not recognize the faces properly in different lighting conditions. Example if I train the system in sunlight conditions and then re-execute the recognition process in some darker area with no sunlight (eg. in tube light conditions) it does not recognize the face correctly and vice-versa. Also sometimes the face is not detected correctly. The system sometimes detects the wall as a face or some non living thing as a face.
Is it that the algorithms used are not efficient enough to be used for real time systems? Do I need to change the algorithms used? Are there any better algorithms which I can try? Or is there any way by which I can continue using the same algorithm but make the system more efficient?
Please reply. Thank you in advance.