Crowd Fund with Aadhaar Verification and Donor Voting System

Crowd Fund with Aadhaar Verification and Donor Voting System

Hareesh B

Associate Professor Department of Computer Applications

SJEC, Mangaluru

Ajay

Department of Computer Application,

Visvesvaraya Technological University, India

Akshatha

Department Computer Application, Visvesvaraya Technological University, India

Abstract – In order to guarantee the authenticity and integrity of fundraising campaigns, this paper proposes a transparent and safe crowd-funding system that uses donor voting and Aadhaar- based identity verification. Users are validated via selfies and Aadhaar cards using optical character recognition (OCR) and facial recognition algorithms. To increase public confidence in campaign execution, a voting mechanism is incorporated to enable verified donors to accept or deny requests for fund withdrawals. Modern open-source technologies, such as Tesser- act OCR for text extraction, Python Flask for backend APIs, MongoDB for data storage, and React for the user interface, are used to describe the system architecture, data flow, and implementation modules. The suggested model gives donors more control over how their money is used, increases accountability, and lowers fraud in online fundraising platforms.

Index Terms – Crowdfunding, Aadhaar Verification, OCR, Flask API, Donor Voting, Fund Disbursement, Face Recognition, Data Integrity.

1. INTRODUCTION

   Crowdfunding has become an effective instrument for rais- ing money for startups, social causes, urgent medical needs, and educational projects. Online crowdfunding platforms have grown in popularity in India due to the wide range of internet access and digital payment infrastructure. But this growth has also brought with it problems like fraud, fraudulent campaigns, and unauthorized use of funds, which destroy platform trust and donor confidence.

   This study suggests a user-verified, transparent, and safe crowdfunding platform which includes donor-controlled voting and Aadhaar-based identity verification in order to solve these problems. Indias national biometric identity system, Aadhaar, offers an accurate method to verify user identities. To verify the true nature of fundraisers, the suggested system applies facial recognition algorithms to compare Aadhaar photos with real-time user selfies and uses

   optical character recognition (OCR) to extract information from uploaded Aadhaar cards.

   In addition to identity verification, the system presents a milestone-based voting model that encourages donors. Every time a fundraiser requests a fund withdrawal, donors are informed and have the option to vote on behalf of or against the payout. By encouraging transparency and shared respon- sibility, this system of government lowers the possibility of financial misuse.

   Modern open-source technologies such as Flask (Python) for the backend APIs, MongoDB for the database, React for the frontend, and machine learning tools like Tesseract OCR

   and OpenCV for Aadhaar verification are used in the plat- forms development. For oversight, identification of fraud, and manual disable in emergencies, the architecture also includes admin and super admin modules.

   With the goal to improve trust in digital fundraising, this work strives to develop a next-generation crowdfunding solution that is suitable for the Indian ecosystem by combining identity assurance, user participation, and data privacy.

2. SYSTEM ARCHITECTURE AND COMPONENTS

   The presentation layer, business logic layer, data access layer, and external service layer are the four main parts of the multi- tier model that types the basis of the suggested systems architecture. Together, these elements provide a safe, transparent crowdfunding process that includes donor- controlled fund distribution and identity verification.

   1. 1. Presentation Layer

      React, a contemporary JavaScript library for creating dynamic and responsive user interfaces, is used to implement this layer,

      which is in control of user interaction. A web- based front-end that offers smooth navigation and real-time updates is how usersincluding administrators, donors, and fundraisers interact with the system. Component-based development, made possible by React, makes the user interface (UI) modular, reusable, and simpler to maintain. This layer makes it possible for:

      • Campaign creation and browsing

      • Aadhaar upload and verification initiation

      • Donor voting dashboard

      • Admin approval/rejection of campaigns

   2. 2. Business Logic Layer

      This layer contains the systems central logic. It manages:

      • Campaign management

      • Voting decision logic and deadlines

      • Validation of Aadhaar details

      • Workflow automation (e.g., notification triggers, mile- stone updates)

        Java Script and Flask (Python) APIs communicate securely to enable these functions.

        Fig. 1: Aadhaar verification flow using OCR and face recog- nition. The extracted text is compared with form input, and selfie is matched with Aadhaar image.

   3. 3. Data Access Layer

      All transactional and verification data are stored and man- aged in a structured manner using MongoDB. Key entities include:

      • User table with roles (fundraiser, donor, admin)

      • Campaign table with milestones, goals, and status

      • Aadhaar verification records

      • Voting logs with timestamps and results

      • Payment and fund transfer history

   4. 4. External Service Layer

      This includes machine learning and third-party services integrated into the system:

      • Tesseract OCR for Aadhaar text extraction

      • OpenCV and DeepFace for face matching

      • Secure payment gateway for donations and disbursement

      • Flask-based APIs for verification logic

   5. System Security and Access Control

      The platform enforces role-based access control (RBAC). Admins and super admins have additional permissions to:

      • View verification status

      • Approve/reject campaigns

      • Override voting outcomes in rare cases

        All user sessions are authenticated using hashed credentials. Aadhaar data is masked in the UI and not stored in raw form for privacy.

        Fig. 2: Layered system architecture for Aadhaar-based crowd- funding platform.

   6. User Roles

      • Fundraiser: Can register, verify Aadhaar, and create campaigns.

      • Donor: Can view campaigns, donate, and vote for fund release.

      • Admin: Reviews and approves/rejects campaigns and monitors activity.

      • Super Admin: Oversees system operations and manages roles and reports.

3. AADHAAR VERIFICATION WORKFLOW

   The Aadhaar verification system forms the backbone of identity authentication in the proposed crowdfunding

   platform. It ensures that fundraisers are genuine individuals by leveraging both textual and facial data verification.

   1. Step 1: Aadhaar Upload

      Users are required to upload a scanned image or a clear photo of the front side of their Aadhaar card. The uploaded image is temporarily stored for analysis via a secured Python Flask API.

   2. Step 2: OCR Text Extraction

      The uploaded Aadhaar image is passed through the Tesser- act OCR engine to extract key details:

      • Full name

      • 12-digit Aadhaar number

      • Date of birth/p>

      • Gender

        These values are then cross-checked with the values entered manually during the registration process. Any mismatch flags the user as unverified.

   3. Step 3: Live Selfie Upload

      Simultaneously, users are prompted to capture and upload a live selfie. This selfie is used for facial comparison with the photo extracted from the Aadhaar card.

   4. Step 4: Face Matching

      Using OpenCV and the DeepFace facial recognition library, the Aadhaar card photo and the live selfie are compared for similarity. A facial similarity score is calculated. If the

      confidence level is above a predefined threshold (e.g., 85%), the face verification is considered successful.

   5. Step 5: Verification Status and Feedback

      Both the OCR match and face match results are used to determine the users Aadhaar verification status:

      • If both checks succeed: Status is set to Verified

      • If either check fails: Status is set to Rejected

      • If confidence is borderline: Status is set to Pending Review

        Admins are notified of failed or pending verifications for manual intervention.

   6. Security and Privacy Measures

   All image uploads and API calls are encrypted using HTTPS. Aadhaar numbers are partially masked (e.g., only last 4 digits shown) on the frontend to preserve privacy. Verification logs are maintained for auditing but images are deleted from the server after verification.

4. DONOR VOTING AND FUND MANAGEMENT

   One of the key innovations of the proposed platform is the integration of a donor-driven voting mechanism for fund disbursement. This feature enhances transparency, ensures accountability, and gives donors more control over how and when funds are released to the campaign organizer.

   1. Fund Disbursement Model

      This system divides the fund release into predetermined milestones, compared to traditional platforms where money is automatically transferred to the campaign owner upon reaching the goal. A request for fund release is made by the fundraiser each time they hit a milestone. The donors are then contacted to review this request.

   2. Donor Voting Mechanism

      The release request is communicated to all campaign donors via email or dashboard notification. They have a short window of time (three to seven days) to vote on behalf of or against the release. Donor IDs and timestamps are used to safely store votes in the database.

      • If a majority (50%) approves the release, the amount is disbursed to the fundraiser.

      • If the majority rejects the request, the disbursement is withheld.

      • If there is a tie or insufficient quorum, the admin reviews the case manually.

   3. Vote Transparency and Logging

      Every vote is recorded in the system, and the campaign dashboard displays a summary. By demonstrating progress, this not only encourages community involvement but also enhances the fundraisers authority.

   4. Fund Flow and Refund Handling

      After approval, money is transferred to the fundraisers registered account through a secure payment gateway.

      Unused funds are refunded to donors if a campaign misses its goal or is detected during verification.

   5. Admin Supervision

      Admins have privileged access to override a voting outcome only under exceptional circumstances such as:

      • Legal intervention

      • Policy violation

      • Fraud detection

   Such actions are logged with justification for auditing. Fig. 4: Donor voting system. Donors vote after fundraiser

   milestone updates. Based on majority decision, funds are released or held for admin action.

5. RESULTS AND EVALUATION

   The proposed system was evaluated based on three key parameters: verification accuracy, user feedback, and system performance. Testing was carried out using a sample dataset of Aadhaar card images and user selfies under various

   lighting and resolution conditions.

   1. Aadhaar Verification Accuracy

      The Aadhaar verification pipeline was tested with 150 different image pairs of Aadhaar cards and corresponding selfies. The following results were observed:

      • OCR Accuracy: Using Tesseract OCR, the system achieved an average text extraction accuracy of 92% for name, DOB, and Aadhaar number fields.

      • Face Match Accuracy: OpenCV combined with Deep- Face achieved 95% accuracy in correctly matching the face from the Aadhaar card with the live selfie, provided the image quality was clear and lighting was consistent.

      • False Acceptance Rate (FAR): Less than 2%, indicating strong resistance to impersonation attacks.

      • False Rejection Rate (FRR): Approximately 4%, mostly due to poor image quality or occlusions.

   2. System Performance

      Performance benchmarks were conducted on a mid-tier cloud VM (4-core CPU, 8GB RAM).

      • Average API Response Time (Flask): 180ms for Aad- haar OCR and face comparison.

      • Frontend Load Time (React): 1.2s average with opti- mized image compression.

      • Database Throughput: Capable of handling 500 concur- rent read/write operations per second without significant latency.

   3. Usability Testing

      A small group of 12 end users (6 fundraisers and 6 donors) tested the system in a pilot run.

      • 100% of users found the UI intuitive and easy to navigate.

      • All testers were able to complete registration, Aadhaar upload, and campaign participation without external help.

      • Donors appreciated the transparent fund control via the voting module.

   4. Limitations

      While the system performed well, a few limitations were noted:

      • Performance of face verification decreases

        significantly under poor lighting or blurry images.

      • Tesseract OCR fails to extract correct data from Aadhaar images with artistic backgrounds or severe glare.

      • Voting requires an active user base; if donors are

        inactive, campaigns may stall unless overridden by the admin.

6. CONCLUSION

This paper suggested a new way to make crowdfunding safe and open by combining Aadhaar-based identity verifi- cation with a voting system that lets donors choose what to do. The proposed solution solves two major problems that current crowdfunding platforms have: fake campaigns and donors not having enough control. It does this through new technology and careful handling of data. The platform makes sure that only verified Indian citizens can start fundraising campaigns by using Optical Character Recognition (OCR) and facial recognition. Flask-based APIs connect the front-end and machine learning back-end modules.MongoDB safely stores transaction and voting logs. The system gives donors more power by letting them vote on how the money is spent, which greatly increases responsibility and public trust.

Comprehensive testing showed that both OCR and facial matching were very accurate, which shows that this kind of system could work in the real world. Also, the modular design makes it easy to add more features and connect it to other services like UPI, blockchain, and Aadhaar eKYC.

Despite these successes, there are still restrictions. Because the system depends on high-quality image inputs, low- resolution or weak illumination can affect facial matching performance. In addition, because of regulatory limitations, eKYC integration with official Aadhaar APIs is still pending.

Future work will include:

• Integration with UIDAI eKYC and biometric authentication

• Blockchain-based transaction ledger for immutable audit trails

• NLP-enabled chatbot to guide user through registration and verification

• Mobile app version with offline upload queue

In short, the proposed Aadhaar-integrated crowdfunding platform uses identity verification, flexibility, and user empowerment to make a safe place for raising money that is hard to fraud. It sets the foundations for Indias and other similar places next-generation social finance systems.

REFERENCES

1. UIDAI Aadhaar API Documentation. [Online]. Avail- able: https://uidai.gov.in

2. Tesseract OCR Optical Character Recognition Engine. [Online]. Available:

   https://github.com/tesseract-ocr

3. S. Rosebrock, Face recognition with OpenCV and Deep Learning,

   PyImageSearch, 2021. [Online]. Available: https://pyimagesearch.com

4. Flask Framework. [Online]. Available: https://flask.palletsprojects.com

5. MongoDB Documentation. [Online]. Available: https://www.mongodb.com/docs

6. React A JavaScript library for building user interfaces. Meta Platforms Inc. [Online]. Available: https://reactjs.org