Exam cheating detection

Nowadays, cheating has become really common among the students in exam halls. You have to take a lot of measures to prevent students from cheating and still the problem remains unsolved. So we are creating a cheating detection camera software that will detect the cheaters during examinations in exam

Project Title

Exam cheating detection

Project Area of Specialization

Artificial Intelligence

Project Summary

Nowadays, cheating has become really common among the students in exam halls. You have to take a lot of measures to prevent students from cheating and still the problem remains unsolved. So we are creating a cheating detection camera software that will detect the cheaters during examinations in exam halls and catch them while cheating. It will use facial recognition libraries and YOLO algorithm to detect the images of the students trying to cheat. It is a hybrid application where the user will have complete access to the front end where they will be able to control everything and check the students caught cheating. This cheating detector will detect cheating by comparing the images of students in normal state and cheating state with a specific threshold we set which will be consider cheating state if it went up than threshold. After placing high definition cameras in class through our software we will monitor the students. Our system will be trained to detect specific cheating actions and mark the students. Through video our system will match images from our data base to detect whether the specific student is cheating or not by marking them. If there score increase our system will sent alert of cheating detected. Our system will mark the student with the information of their class. The caught students will have their cheating images stored with them. Basically, our product is designed to detect cheating in exams. This is a really common issue in all institutes to stop cheating. Our software will help stop cheating. This is an exam cheating detector software, So we are creating a cheating detection camera software that will detect the cheaters during examinations in exam halls and catch them while cheating. It will use facial recognition libraries and YOLO algorithm to detect the images of the students trying to cheat

Project Objectives

The primary objective of this project will be to detect cheating of students in exam so the invigilator can take actions against the particular student. The cctv cameras installed in exam hall will be capturing images of the students and will be comparing them to our reference images that we will include in our dataset. The student looking in some specific poses will be detected by camera and his/her folder will be created in desktop app with their images from where the invigilator can see the particular student cheating. If the same student is caught by the camera more than 5 to 6 times and the threshold goes higher than assigned value, then we can say that he/she was cheating and invigilator can prove it by pictures present in the cheating folder. Usually system uses 30fps or 60fps. But we will take 1 or 2 frame rate per second because for 30fps we will have to use heavy systems and will be costly and not any can afford it easily. For cheating a student requires at least 2, 3 seconds that’s the minimum for a student to cheat so 1, 2 frames per second will be enough for us to detect cheating.  

Project Implementation Method

We have collected our data set ourselves by visiting various exam classrooms and halls and took the pictures of student giving exams in nomal state which will be used to compare the cheating state images with. We have also collected cheating state poses which are used in cheating. They are different cheating poses which are to be considered as cheating state. The algorithm which we are using to detect cheating is YOLO v5 because it is the most accurate and efficient. We train our Yolo model according to our requirements to detect cheating. We have desgined a frontend where examiner will be able to operate the software. Once, the software starts monitoring, the cameras placed in classrooms will be rendering video constantly. The video will be send to our system where our system will extract different images of students and send them to model. Our model will than detect suspicious behavior by image comparing. Once some suspicious behavior of one student is detected, its image will be processed and the face will be extracted and image of cheating state will be compared with normal state and then the threshold will be checked, and if the cheating threshold is above than normal threshold assigned by us. The cheating proof will appear in cheating folder in our desktop app where the user will be able to access it. Once the cheating state is confirmed and cheating folder is created, it will display an alert message with beep sound to notify the user so he can take actions however he/she wants with the proof they have.Our software is not responsible to for detection small cheating objects like paper chits, usbs, any cheating material or mobile phones. It detects some poses that are required by the students to perform while they cheat.

Benefits of the Project

•No staff will be required in the class rooms.

•Fair Examinations will be held.

•Cheating rate will be reduced.

•Cameras will be observing each second so students will fear to do cheating.

•Discipline will be maintained.

•Only onetime cost is required to set up the system.

•Teachers time will be saved.

Honest and fair examination means clear and efficent results and proper talent hunt.

We can differ between a students who scored by his / her hardword and the one who scored marks by cheating.

This will benefit the education system of Pakistan.

This will help produce pure talented and honest student to go forward and grab oppurtunities in future.

Technical Details of Final Deliverable

Our system is custom built which is designed according to needs and wants of client. The type of processing our system is responsible for is online processing and analytical reporting. The major application components of our system include camera, desktop PC and our model. Our current system collects and manage videos and images of student being monitored in exam. Other than that, our system uses layered architecture. The main programming language we are using is python. The hardware platform that supports our current system is Desktop/PC. We are using firebase as database to support our current systems. Our system has no user interface. We are going to be using LAN network architecture in this project. Lastly, the system is hosted by an External Data center.

We have three major application components
Camera

Desktop

Trained Model

The camera is connected to desktop software where our trained model is also present. The camera sends video processed to the desktop. The system then extracts images of students and then the model comes in. Here cheating will be detected according to our algo and model and if cheating is found, the user will be notified.

The high level technology used to support our software is minimum 8gb ram or16gb ram. HD camera for better video processing results. Other than that we are required 4 or 8 GB minimum GPU to run the whole system smoothly. The processor should also be fast minimum i3 or i5.

Final Deliverable of the Project

Software System

Core Industry

IT

Other Industries

Education

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Quality Education

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com