Smart Invigilator

Project Title

Smart Invigilator

Project Area of Specialization Computer ScienceProject Summary

In an examination hall, invigilators are instructed to perform a variety of duties, the most important duty is to maintain discipline and make sure the examinees do not cheat. In this age where computers are now able to perceive and analyze their environment with a high degree of accuracy, we can use this advancement to automate the process of invigilation.

Smart Invigilator will be used by the person in charge of the examination hall and activated at the start of the exam. Cameras would be installed in the examination hall and they will stream the video to the system at backend, which will analyze the stream, frame by frame, and it will detect the examinees in the frame. Furthermore, it will monitor the movement of examinees in the video stream and detect suspicious activities. Finally, the suspicious movements will be reported to the head invigilator on his/her mobile.

Project Objectives

The objectives are listed below:

1. To make invigilation more reliable.
2. To make invigilation human less.

Project Implementation Method

The implementation consists of the following steps:

• Survey of Existing System
• Data Collection
• Model Design
• Testing

Each process is discussed below in detail.

  Survey of Existing System

In this phase detailed literature review will be conducted on face detection as well as movement detection to gather information about the existing work. Furthermore, the aim would be to find existing techniques and technologies that may help in the project. Literature with the latest publication (2010-2021) is planned to be concatenated but classes are also being explored.

  Data Collection

Data Collection is considered as the most challenging task in conducting a research. This research is based on real time face detection and motion recognition in a controlled environment. For gathering data, the humanoid action frames would be captured that are suspicious and non-suspicious during an exam. Face detection algorithm would be used to detect face and extract the region of interest in rectangular bounding box. All extracted images will be converted to grayscale and resized equally. Afterwards, the dataset would be used to train the system model.

  Model Design

At initial stage, the Viola Jones algorithm would run on the dataset to identify the suspicious and non-suspicious haar-like features and creating a boosted classifier cascade. Afterwards, the Cascade Classifier would be used for the recognition of suspicious features based on the generated haar-cascade feature list. If a suspicious feature is recognized, the head invigilator would be notified.

  Testing

The system would be thoroughly tested for inconsistencies, the testing would, first, be conducted on test data i.e., recorded video, and when the testing on recorded video is successful, the system would then be tested using a live video feed. The reliability of the system would be evaluated by the accuracy of face detection and motion recognition.

Benefits of the Project

The benefits of the project are listed below:

• Financial benefits (Only few invigilators required)
• Proof of suspicious activity
• Effortless and reliable invigilation

Technical Details of Final Deliverable

• The final deliverable would be a hardware & software integrated system.
• The frames taken from camera would be used as an input to the system at backend.
• The system would perform face detection and motion recognition in the video stream received from camera.
• The programming language would be Python.
• The mobile app development framework would be React Native.
• The notification would be delivered via Firebase Cloud Messaging (FCM).
• The library under use would be OpenCV for building the model.

Final Deliverable of the Project HW/SW integrated systemCore Industry ITOther Industries Education Core Technology Artificial Intelligence(AI)Other Technologies OthersSustainable Development Goals Quality EducationRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	74900
Camera	Equipment	4	10000	40000
4 Channel DVR	Equipment	1	10000	10000
Tripod	Equipment	4	2500	10000
Cables & Connectors	Equipment	1	5000	5000
CV In Python! Face Detection & Image Processing (Udemy Discounted)	Miscellaneous	1	2200	2200
Python DIP From Ground Up™ (Udemy Discounted)	Miscellaneous	1	2200	2200
Complete Guide to TensorFlow for DL with Python (Udemy Discounted)	Miscellaneous	1	2300	2300
The Complete React Native + Hooks Course (Udemy Discounted)	Miscellaneous	1	3200	3200