PROJECT OVERVIEW

A. DEFINING THE PROJECT

🔥 Project Summary

Customer: Polis Diraja Malaysia (PDRM) (Mock)

Project name: iCrime

Team Members:

Muhammad Azri Bin Azmi
Aiman Iskandar Bin Mohd Zaidi
Ariff Rahimin Bin Mohamed Norazman
Muhammad Haziq Izzuddin Bin Mohd Junaidy

😁 Objectives:

Identify criminals by using face recognition.
Reduce workload for police.
Improve law enforcement.

B. PLANNING THE PROJECT

💻 Project Management Life-Cycle

Figure 1: Gantt Chart

Figure 2: Work Breakdown Structure

💹 Risk Identification Chart (Quality, Cost, Time)

Control Element	What is likely to go wrong?	How and when will I know?	What will I do about it?
Quality	Faces could not detect accurately.	Cameras detect wrong criminals faces.	Modify the algorithm like threshold to detect the faces.
Cost	Overexpensive equipments such as cameras.	Funds are spend to hardware only.	Find new suppliers to cover the costs.
Time	Time consuming to collect and train the data.	Training process took so long and sometimes have an error.	Train the data by using GPU instead of CPU or using cloud computing like Google Colab.

📗 Responsibility Assignment Matrices (RAM) :

Figure 3: Responsibility Assignment Matrices

Name	Roles
Muhammad Azri Bin Azmi	Project Manager
Aiman Iskandar Bin Mohd Zaidi	Design Technician
Ariff Rahimin Bin Mohamed Norazman	Lead Engineer
Muhammad Haziq Izzuddin Bin Mohd Junaidy	Senior Project Director

📌 Project Planning Summary:

Modules/Components	Budget (RM)	Schedule	Responsibility
Criminal Database	500,000.00	30 December 2020 - 9 January 2021	Collect Data, Label Data
Criminal Face Recognition	1,200,000.00	17 January 2021 - 30 January 2021	Train model, Test model
Map Tracking	120,000.00	10 January 2021 - 16 January 2021	Connect system with GPS

C. IMPLEMENTING THE PROJECT PLAN

📬 Deliverables:

Planning Deliverables
Analysis Deliverables
Design Deliverables
Implementation Deliverables

📂 Tasks and Estimated Costs

Task	Estimated Costs(RM)	Notes
Acquisition	5,272,630.00	-
Design	391,110.00	System Design, Data Collection
Restoration	1,328,972.00	Camera Installation, System Installation, Maintenance

📆 Milestone Chart

Milestone	Scheduled Completion	Actual Completion
Planning	28 November 2020	27 November 2020
Analysis	30 December 2020	21 December 2020
Design	29 January 2021	30 January 2021
Implementation	28 February 2021	26 February 2021

D. EXECUTING THE PROJECT

💡 Project Design and coding

Flow Diagram

Figure 4: Flow Diagram

😁 CODING:

Face Recognition
Command: pip install -r requirement.txt

Use this command to install the required packages.

Figure 5: Bounding Box

For every face that are detected, a box will be placed based on the face locations. It will crop the face, and apply the model that we train, to recognize the face.

Figure 6: Filter algorithm

This algorithm filters raw images and select images that have face, and remove images that have no face or bad to detect face.

Figure 7: Training model

Images with each label(their folder), are being extracted and trained with MTCNN.

PATH PLANNING

After camera can recognize the criminal face, this algorithm will execute to search the shortest path from police station to criminal(camera)
In this project, we use Breadth First Search algorithm for path searching