2. Agenda
• Company Profile
• Project Definition
• Project Methodology
• Project Scope
• Features
• Tools and Technology
• System Flow Diagram
• Database Design
• Data Flow Diagrams
• Screenshots
• Conclusion
• Future Enhancement
3. Company Profile
• Our client is The Maharaja Sayajirao University which is has
numerous students which makes managing them important.
• At present, attendance, marking involves manual attendance on
the paper sheet by professors and teachers, but it is a very time-
consuming process and chances of proxy are also an issue that
arises in such type of attendance marking.
4. Project Definition
Asistencia is a system that takes attendance automatically,
eliminating time and efforts taken during the traditional
approach.
It focuses on attendance of students and faculties
D.Pg: 5
5. Project Methodology
• Research comprises creative and systematic work undertaken to
increase the stock of knowledge, including knowledge of
humans, culture and society, and the use of this stock of
knowledge to devise new applications.
• It is used to establish or confirm facts, reaffirm the results of
previous work, solve new or existing problems, support
theorems, or develop new theories.
• A research project is an expansion on past work in the field.
D.Pg : 7
6. Project Scope
• Provides facility for the automated attendance of students.
• Uses live face recognition to recognize each individual and
mark their attendance automatically.
• Utilizes video and image processing to provide inputs to
the system.
• Facility of knowing the percentage of presence during a
month.
• Facility of marking manual attendance.
• Notification via email if there is a lack of attendance.
D.Pg : 9
7. Features
• Authorized by administrator.
• Registration of student details and faces.
• Automatically captures 100 images at the time of registration
• Creation of student’s folder where images are stored is created
automatically.
• Face are recognized matching with registered image
• Unknown persons are identified
• Excel sheet and sheet is created automatically if not created
• Reports are generated as an when required
• E-mail notification
41. Conclusion
• While using K-NN there was sharp fluctuation in the prediction rates
which lead to changing performance parameters constantly for every
newer condition.
• K-NN kept performing worst as the size of the dataset increased thus
concluding that it’s a best fit for a small dataset.
• We found that the accuracy and other performance parameters depend
more on the no of epochs, so SVM was much better when compared
with the K-NN.
D.Pg : 60
42. Comparison-1
General Comparison
D.Pg : 62
Dlib’s model with 68 land markpoints OpenCV Haar-Cascade for face detection
• The pre-trained facial landmark
detector inside the Dlib library is used
to estimate the location of 68 (x, y)-
coordinates that map to facial
structures on the face.
• OpenCV comes with a trainer as well
as detector. If the detection window
fails the first stage, discard it and don't
consider remaining features on it.
• If it passes, apply the second stage of
features and continue the process. The
window which passes all stages is a
face region.
43. Images
per person
Accuracy with i7 4th
Generation
Accuracy with i3 5th
Generation
Accuracy with i5 4th
Generation
SVM K-NN SVM K-NN SVM K-NN
50 Images 17.4 Min 15.3 Min 19.3 Min 16.1 Min 17 Min 14.2 Min
100
Images
31.8 Min 28.5 Min 38 Min 34 Min 34.6 Min 25.5 Min
Comparison-2
Processing Power
D.Pg : 61
47. Average SVM K-nearest neighbor
1.Accuracy of Recognition 90 % 60 %
2.Processing time for encoding
and training(in Minutes with 30
classes)
26.1 22
Comparison-6
Overall Comparison
D.Pg : 60
48. Future Enhancement
• For quicker performance processing can done using Graphical
Processing Unit (GPU).
• Further this technology can be implemented by 3D based face
recognition.
• Data Storage can be made server-based.
• System can be integrated with multiple cameras at the same time.
D.Pg : 65