Skip to content

Latest commit

 

History

History
64 lines (54 loc) · 3.19 KB

README.md

File metadata and controls

64 lines (54 loc) · 3.19 KB
Raw

Project Description

Godseye is an AI based project working on the principle of Computer Vision.
Using digital images and live feed it detects and recognizes human face and mark their attendance.
The sole purpose of this model is to detect a person and add their details separately in Attendance list.

Novelty

  • Instead of working on a single face, our model can recognize multiple face in one frame.
  • Our model not only recognizes a face but also mark attendance of that recognized person.
  • Our model update Attendance list in real-time (as soon as it recognizes a person).

Real-time Usage

  • Firstly, we open the webcam to take images of people present in front of the camera.
  • Then we check each of them if they exist in our data or not. If not, they'll be called as 'Unknown'.
  • After checking each of them with our data, we check that if they already marked as present or not. If not, we take that person's Registration no., Name & Entry time and mark them present.

Hardware & Software Requirements

  1. Hardware: Desktop or laptop with a webcam installed
  2. Minimum Specs: 4gb RAM and 80gb HDD dual core processor
  3. OS: macOS or Linux (Windows not officially support 'face_ recognition' library, but it might work)
  4. Programming Language: Python 2.7 or Python 3.3+
  5. Application: 'Spreadsheet' in macOS or 'LibreOffice Calc' in Linux

Screenshot of Project

Testing

  1. We started our project with testing face recognition in different images by Image Recognition.
  2. Then we developed an algorithm to find the location and encodigs of multiple faces in one frame.
  3. After that we wrote the code by importing all the required libraries & our developed algorithms.
  4. After recognizing the face, we are checking that person's details and finding it's accuracy.
  5. At last we made .csv file in which we are storing the attendance of the recognized people.

Result & Discussion

  • This method can detect multiple face in one frame and can be easily used in a classroom or in an office.
  • This system helps us to achieve desired results with better accuracy and less time consumption.
  • The precision or the accuracy of face recognition of our model is almost more than 90%.

Conclusion

Thus, the aim of this model is to capture the video of the students/colleagues, convert it into frames, relate it with the dataset to ensure their presence or absence, mark attendance to the particular student/colleagues to maintain the record. The Automated Classroom Attendance System helps in increasing the accuracy and speed ultimately achieve the high-precision real-time attendance to meet the need for automatic classroom evaluation.

Steps to Run

  • Each .py file has a specific work to perform (Commented in the starting of that file) and videoCode[final.py]
  1. Fork this repo
  2. Clone the forked repo to your local system
  3. Install the following libraries: (in Linux or macOS)
    1. cv2
    2. face_recogniton
    3. os
    4. math
    5. numpy
    6. datetime
  4. Run the code

Support

If you like this project, don't forget to give it a ⭐