AI Project - Aainaa0/Hand-written-Text-Recognition GitHub Wiki
Hand-written Text Recognition (HTR) is a special type of OCR which deals with the ability of computer to recognize the letters written by human. HTR has become a vital part of robotics vision for its application in enormous sectors of life like smart education, autonomous driving, purchasing process, digitizing paper based process and many other areas. It has a lot of prominent applications such as National ID number recognition, post office automation, automatic plate recognition system for vehicles, online banking, parking lot management system, and many more.
In this project, we will focus on the hand-written text recognition module.
- Tan Yi Wen
- Aainaa Nabilah binti Rohaizad
- Nur Alis Sophia binti Suhaimi
- Muhammad Nur Irfan bin Azman
- To develop a tool that will scan the inputted images and printed out extracted text as the output.
- To develop a tool that can predict the handwritten character with above 90% probability.
- To develop an Artificial Intelligence system that are able to detect user's handwritten characters or words.
- To construct a tool that takes an image from it as an input and extracts characters (alphabets, digits, symbols).
The project management life cycle is represented and documented in the form of Gantt Chart. A Gantt chart is a project management tool assisting in the planning, showing start and end dates, as well as dependencies, scheduling and deadlines, including how much of the task is completed per stage and who is the task owner.
The WBS is the Work Breakdown Structure. It is a hierarchical representation of the elements (tasks) that comprise a project. WBS of our project is shown below:
Project management WBS is shown in the Gantt Chart below:
| Control Element | What is likely to go wrong? | How and when will I know? | What will I do about it? |
|---|---|---|---|
| Quality | The system unable to detect messy hand-written text | How: The result output is wrong. When: During testing phase. |
Put more effort in collecting data and increase the data set every day until a satisfactory result is determined. |
| Cost | Project runs over budget | How: Not enough costs to continue. the project. When: During project implementation phase. |
After coming up with an agreed budget with the client, we must manage the expenditure carefully, Expenditure must be controlled to ensure that the project is completed without overbudget. |
| Time | Project do not follow the milestone. | How: The progress of project stuck in certain task. When: During project implementation phase. |
Assign other team member to help with that certain task. |
| Roles | Person In Charge | Responsibility |
|---|---|---|
| Project Manager | Tan Yi Wen | Project managers play the lead role in planning, executing, monitoring, controlling and closing projects. Project managers are accountable for the entire project scope, project team, resources, and the success or failure of the project. |
| Risk Manager | Alis | The role of a Risk Manager is to communicate risk policies and processes for Lumineux Sdn. Bhd.. They provide hands-on development of risk models involving market, credit and operational risk, assure controls are operating effectively, and provide research and analytical support. |
| Administrative Manager | Alis | Administrative Managers supervise administrative team members and ensure that daily office operations are performed in a seamless and efficient manner. Administrator manager needs to list all possible content that can be displayed in the document form after each process in the hand-written text recognition project. |
| Financial Analyst | Irfan | Financial analysts are responsible for supporting administrative managers by managing and tracking the costs of hand-written text recognition projects, preparing budget documents. For example, Special Project Reports and OSI Budget Change Proposals, reviewing budget/contract expenditures, and collecting and our reporting financial metrics. This includes reconciling the accounting and Work Breakdown Structure cost management processes and developing financial management policies and procedures. |
| Quality Manager | Irfan | Quality Manager provides insight into the hand-written text recognition project and contractor methods of doing business by reviewing process and product activities for adherence to standards and plans. Quality Manager have to Understand customer needs and requirements to develop effective quality control processes |
| Technical Manager | Aainaa | Technical Manager is responsible for making key decisions for technical issues related to the company after considering everything from cost to quality of hardware or software. The technical manager must provide guidance for all members of the team when it comes to designing, implementing, and updating software. |
| System Enigineer | Aainaa | System Engineer has responsibility in managing and monitoring all installed systems and infrastructure, Installing, configuring, testing and maintaining operating systems, application software and system management tools, as well as ensuring the highest levels of systems and infrastructure availability. |
| Implementation Manager | Aainaa | Implementation project manager is responsible for managing a team of implementation specialists, who develop new software or configure new hardware for the client. They ensure teams are on task and team deadlines are met. |
| Test Manager | Irfan | Test manager is responsible for the planning, coordination, and control of various test activities. They first create a test process and test strategy that analyzes which test methods and tools are suitable and used for test activities. |
| Configuration Manager | Tan Yi Wen | Configuration Manager has responsibility to Identify, organize, and control software and hardware configuration changes. Identify and coordinate processes for version management, system build, backup and recovery, archiving, and change management. Support and facilitate auditing and reporting. |
- Python code
- Project documentation
- Best trained model
| Task | Estimated Costs (RM) | Notes |
|---|---|---|
| Software | 5,000.00 | |
| Scanner | 6,000.00 | |
| Computer | 12,000.00 | |
| Database | 1,800.00 | |
| Travel Expenses | 50,000.00 | |
| Rentals | 160,000.00 | |
| Wages | 12,000.00 | |
| Advertising | 150,000.00 | |
| Maintenance | 16,000.00 | |
| Insurance | 8,000.00 | |
| Building Rent | 4,500.00 | |
| IT Services | 2,000.00 | |
| Data Collection | 500,000.00 | |
| Conseptual Design | 6,000.00 | |
| Preliminary Design | 15,000.00 | |
| Final Design | 1,000.00 | |
| Construction Labor | 12,000.00 | |
| Construction Supervision | 10,000.00 | |
| Materials | 50,000.00 | |
| Permits | 300.00 | |
| Surveys | 2,000.00 | |
| Architectual Design | 18,000.00 | Overall design of the project |
| Rentals | 10,000.00 | Rentals used such as hardware rent and building rent. |
| Software | 1,000.00 | Software used to develop the project. |
| Indirect Costs | 128,000.00 | |
| TOTAL | 1,180,600.00 |
| Milestone | Scheduled Completion | Actual Completion |
|---|---|---|
| Planning and Analysis | 5/11/2020 | 4/11/2020 |
| Design and Implementation | 24/12/2020 | 24/12/2020 |
| Testing | 11/1/2020 | 10/1/2020 |
| Requirement and Maintenance | 20/1/2020 | 20/1/2020 |
> python main.py
Init with stored values from ../model/snapshot-38
Recognized: "house"
Probability: 0.96177125
1. Customer Acceptance/Project Completion Form [source]
2. Lessons Learned Document [source]
Hand-written Text Recognition
The Hand-written Text Recognition (HTR) System developed by Lumineux Sdn. Bhd. was intended:
- To develop an Artificial Intelligence system that are able to detect user's handwritten characters or words.
- To construct a tool that takes an image from it as an input and extracts characters (alphabets, digits, symbols).
- To develop a tool which can scan the inputted handwritten images and printed out extracted text as the output.
This Hand-written Text Recognition (HTR) System is developed successfully which is able to scan and detect the characters or words correctly.
The project members stick to the plan as in Gantt Chart and all progress are on track which fit to the original schedule.
The actual budget is RM 1,079,440.00, which is RM 101,160 less than the estimate budget costs.
### Tasks and Estimated Costs
| Task | Estimated Costs (RM) | Actual Costs (RM) |
|---|---|---|
| Software | 5,000.00 | 4,500.00 |
| Scanner | 6,000.00 | 6,000.00 |
| Computer | 12,000.00 | 12,000.00 |
| Database | 1,800.00 | 1,500.00 |
| Travel Expenses | 50,000.00 | 47,620.00 |
| Rentals | 160,000.00 | 150,000.00 |
| Wages | 12,000.00 | 10,000.00 |
| Advertising | 150,000.00 | 150,000.00 |
| Maintenance | 16,000.00 | 12,500.00 |
| Insurance | 8,000.00 | 8,000.00 |
| Building Rent | 4,500.00 | 4,500.00 |
| IT Services | 2,000.00 | 1,800.00 |
| Data Collection | 500,000.00 | 429,020.00 |
| Conseptual Design | 6,000.00 | 6,000.00 |
| Preliminary Design | 15,000.00 | 14,000.00 |
| Final Design | 1,000.00 | 1,000.00 |
| Construction Labor | 12,000.00 | 11,000.00 |
| Construction Supervision | 10,000.00 | 10,000.00 |
| Materials | 50,000.00 | 50,000.00 |
| Permits | 300.00 | 300.00 |
| Surveys | 2,000.00 | 1,700.00 |
| Architectual Design | 18,000.00 | 18,000.00 |
| Rentals | 10,000.00 | 10,000.00 |
| Software | 1,000.00 | 1,000.00 |
| Indirect Costs | 128,000.00 | 119,000.00 |
| TOTAL | 1,180,600.00 | 1,079,440.00 |
The project scope of this project is accomplished as the Hand-written Text Recognition (HTR) System which are target to be used in many sectors and applications. Example of sectors are sectors of life like smart education, autonomous driving, purchasing process, digitizing paper based process and many other areas. Example of prominent applications are National ID number recognition, post office automation, automatic plate recognition system for vehicles, online banking, parking lot management system, and many more. The requirement for finishing this project is to have an expert knowledge in neural networking and programming in python.
Lumineux Sdn. Bhd. had successfully achieve the objectives initializes since the beginning of the project. The Hand-written Text Recognition (HTR) System developed was intended to develop an Artificial Intelligence system that are able to detect user's handwritten characters or words and printed out extracted text as the output. In this project,Deep learning is used to identify and read the words or characters on the image of hand-written text.
The success criteria listed in the project scope management is to complete the project within 4 months, and the estimated budget cost is RM 1,180,600.00 . The project should meet all customer requirements specified in the project scope management. The project manager should ensure that after creating and well-controlled Gantt charts in the process, all team members stay on track to achieve scope goals.
Below is the link to our presentation video:
