01 Project Synopsis
₹49
One-time Payment
- PDF & Word Formats
- Project Introduction & Objectives
- Problem Statement & Methodology
- System Architecture Diagram included
- Structured for Review & Customization
M.E. Project Report
M.E. Hand Gesture Recognition Final Year Project Report
Get structured M.E. Hand Gesture Recognition Final Year Project Report Final Year Project Report and Documentation with project objectives, methodology, system design, diagrams, implementation details, testing and complete project explanations. Suitable as a learning, documentation and project-presentation resource for students working on related final year projects.
Need a customized report? Chat on WhatsAppChoose the Report Package That Fits You Best
Simple pricing. Instant access. Every package includes PDF & Word format.
02 Detailed Project Report
₹99
One-time Payment
- PDF & Word Formats
- Structured Project Chapters
- UML / DFD / Technical Diagrams
- Methodology, Implementation & Testing
- Structured chapters & technical diagrams
03 Customized Project Report
₹149
One-time Payment
- PDF & Word both included
- Customized to your college documentation format
- Project-specific customized documentation
- Structured chapters & technical diagrams
- Delivery within 24-48 hours
04 Customized Plagiarism-Free Report
₹299
One-time Payment
- PDF & Word both included
- Plagiarism-free project content
- Original & project-specific writing
- Clear, natural & readable documentation
- Delivery within 24-48 hours
-
—
Abstract
-
—
Table of Content
-
01
Introduction
-
02
Problem Statement
-
03
Existing System
-
04
Proposed System
-
05
Objectives
-
06
System Architecture
-
07
Major Functional Modules
-
08
Hardware Requirements
-
09
Software Requirements
-
10
Future Enhancement
-
11
Conclusion
-
12
References
-
—
Abstract
-
—
Table of Content
-
01
Chapter 1 — Introduction
- 1.1 Background of the Project
- 1.2 Problem Statement
- 1.3 Objectives of the System
- 1.4 Scope of the Project
- 1.5 Existing System Overview
- 1.6 Proposed System Overview
- 1.7 Technologies Used
- 1.8 Overview
- 1.9 Limitations
- 1.10 Advantages
- 1.11 Disadvantages
-
02
Chapter 2 — Literature Review / System Study
- 2.1 Introduction
- 2.2 Review of Similar Systems
- 2.3 Comparative Analysis
- 2.4 Software Development Models
-
03
Chapter 3 — System Analysis
- 3.1 Introduction
- 3.2 Functional Requirements
- 3.3 Non-Functional Requirements
- 3.4 User Requirements
- 3.5 Feasibility Study
- 3.5.1 Technical Feasibility
- 3.5.2 Economic Feasibility
- 3.5.3 Operational Feasibility
- 3.6 System Architecture
- 3.7 Data Flow Diagram (Level 0)
- 3.8 Data Flow Diagram (Level 1)
- 3.9 Data Flow Diagram (Level 2)
-
04
Chapter 4 — System Design
- 4.1 Introduction
- 4.2 Use Case Diagram
- 4.3 Class Diagram
- 4.4 Sequence Diagram
- 4.5 Activity Diagram
- 4.6 ER Diagram
- 4.7 Database Schema
- 4.8 Table Structures
-
05
Chapter 5 — System Implementation
- 5.1 Introduction
- 5.2 Development Environment
- 5.3 Tools and Technologies Used
- 5.4 Hardware Requirements
- 5.5 Software Requirements
- 5.6 Module-wise Implementation
-
06
Chapter 6 — Testing
- 6.1 Introduction
- 6.2 Testing Strategy
- 6.3 Unit Testing
- 6.4 Integration Testing
- 6.5 System Testing
- 6.6 Test Cases
- 6.7 Bug Reports
-
07
Chapter 7 — Results and Discussion
- All screenshots
-
08
Chapter 8 — Conclusion and Future Enhancements
- 8.1 Conclusion
- 8.2 Future Enhancements
-
09
Chapter 9 — References
Project's Overview
Hand Gesture Recognition System Using Python and Machine Learning is a complete Django-based computer vision and machine learning project designed to detect and classify hand gestures through a live webcam or uploaded images.
The system uses OpenCV for image processing, MediaPipe Hands for detecting 21 hand landmarks, and scikit-learn machine learning algorithms such as Random Forest, SVM, KNN, and MLP for gesture classification. The application can work with a rule-based recognition engine even before a custom ML model is trained, making it suitable for academic demonstrations and progressive model development.
Gestura includes approximately 50 predefined hand gesture categories, live webcam detection, uploaded-image recognition, confidence scores, top alternative predictions, hand quality analysis, temporal smoothing, recognition history, user feedback, contact management, dataset management, ML model training, model activation, reports, and separate Admin and User panels.
The project is suitable for B.Tech, M.Tech, BCA, MCA, BE, ME, BSc, MSc, Artificial Intelligence, Machine Learning, Data Science, Computer Vision, and Computer Science students looking for a practical final year project, major project, or minor project involving hand gesture recognition, image processing, computer vision, machine learning, MediaPipe, OpenCV, Django, and real-time webcam detection.
Login Credentials
| Role | Username | Password |
|---|---|---|
| Admin | admin |
admin123 |
| User | None by default | Register from application |
The default admin is created through:
python manage.py create_default_admin
Change the default admin password before using the application in a live environment.
Related & Popular Reports
View all
Admin Dashboard
- View total users
- View total gesture classes
- View total dataset images
- View recognition count
- View active ML model
- View algorithm
- View training accuracy
- View testing accuracy
- View model status
- View unread contact messages
- View feedback count
- View average feedback rating
- View recent recognition activity
- View recent users
- View recent feedback
- Quick access to gestures, dataset, ML, users, history and contact
Gesture Management
- View gesture list
- View gesture label
- View name
- View description
- View image count
- View active/inactive status
- Search gestures
- Filter active/inactive
- Add new gesture
- Edit gesture
- Add class label
- Upload sample image
- Enable gesture
- Disable gesture
- Delete gesture
- Remove associated dataset images when required
Dataset Management
- View uploaded images
- Filter by gesture
- Filter by dataset split
- View total image count
- View valid image count
- View train image count
- View test image count
- Upload multiple images
- Assign images to gesture category
- Upload ZIP dataset
- Use gesture folders inside ZIP
- Reassign image to another gesture
- View train set
- View test set
- Split dataset using train/test ratio
- Remove invalid images
- Remove samples where no hand is detected
- Delete individual image
- Cache MediaPipe landmarks during upload
ML Model Management
- Train machine-learning model
- Upload pre-trained
.joblibmodel - Upload
.pklmodel - Dataset-based model training
- Default one-click training
- Retrain active algorithm
- Save trained models under
media/models/ - Load model
- Activate selected model
- Switch model version
- Delete inactive model
- View training status
- View training accuracy
- View testing accuracy
- Preview prediction
- Upload test image
- Inspect prediction payload
- View F1 score when available
- View training logs
Available ML Algorithms
- Random Forest
- Support Vector Machine (SVM)
- K-Nearest Neighbors (KNN)
- Multi-Layer Perceptron (MLP)
User Management
- View registered users
- Search by username
- Search by email
- Search by name
- Filter active/inactive users
- View user details
- View profile
- View recent recognitions
- View feedback
- Activate user
- Deactivate user
- Delete user
Recognition History Management
- View all recognition records
- View user
- View predicted gesture
- View confidence
- View status
- View engine
- View timestamp
- Filter by user
- Filter by recognition status
- Filter by predicted gesture
- Mark prediction correct
- Mark prediction incorrect
- Delete record
- Bulk-delete records
Feedback Management
- View all user feedback
- View subject
- View rating
- View message
- Delete feedback
Contact Inbox
- View contact messages
- View sender
- View subject
- View message
- View timestamp
- Mark as read
- Delete contact message
Reports
- Total users
- Total gestures
- Dataset image totals
- Recognition attempt totals
- Gesture-wise recognition count
- Most predicted gestures
- Correct prediction count
- Incorrect prediction count
- Unmarked predictions
- Breakdown by recognition status
- Breakdown by engine
- Images per gesture
- Average confidence
- Average feedback rating
Admin Account
- Update admin profile
- Change password
- Logout
The Admin Console manages gestures, datasets, ML models, users, recognition history, feedback, contact messages, reports, and administrative settings.
Authentication
- User registration
- User login
- User logout
- Forgot password
- Password reset
- Session-based authentication
- Profile management
- Change password
- Avatar upload
User Dashboard
- Welcome dashboard
- Gesture count
- Recognition attempts count
- Successful recognition count
- Average confidence
- Quick links
- Recent recognition history
- Active ML model status
- Rule-based engine status
Gesture Catalog
- View active gestures
- Search gesture by name
- Search gesture by description
- View gesture detail
- View gesture class label
- View sample image
- Open gesture in webcam recognition
- Open gesture in upload recognition
Live Webcam Detection
- Start webcam
- Stop webcam
- Browser
getUserMedia - Continuous frame prediction
- Approximate 450 ms live prediction cycle
- Hand landmark overlay
- Draw hand skeleton on video
- Show predicted gesture
- Show confidence percentage
- Show confidence level
- Show recognition engine
- Show latency
- Show hand quality score
- Show prediction stability
- Show alternative predictions
- Show advice when quality is poor
- Save stable recognition results
- Temporal smoothing for live prediction
Upload Image Recognition
- Upload JPEG/PNG image
- Preview image before submission
- Run landmark detection
- Run ML/rule-based prediction
- Generate annotated result
- Save overlay image
- Show recognition report
Recognition Results
- View detected gesture
- View gesture description
- View confidence score
- View engine used
- View latency
- View alternative predictions
- View recognition advice
- Retry using upload
- Retry using webcam
Recognition History
- View personal recognition records
- Filter by success
- Filter by low confidence
- Filter by no hand
- Filter by unknown
- Delete single recognition record
- Clear all personal history
Feedback
- Submit feedback
- Add subject
- Add message
- Add rating from 1 to 5
- View submitted feedback
- Edit feedback
- Delete feedback
Contact
- Send contact message
- Add name
- Add email
- Add subject
- Add message
- View own messages
- View read status
- Delete contact message
The documented user panel includes authentication, profile, dashboard, gesture catalog, webcam detection, image upload recognition, result reporting, recognition history, feedback, and contact modules.
Machine Learning Features
Hand Landmark Detection
The system uses MediaPipe Hands to detect:
- 21 hand landmarks
- Wrist position
- Finger joints
- Finger tips
- Hand geometry
Feature Extraction
Each hand is normalized into a:
42-dimensional feature vector
using x/y coordinates from 21 landmarks.
Normalization includes:
- Wrist-origin transformation
- Palm-scale normalization
- Hand-position independence
- Size normalization
Quality Analysis
The system calculates quality using:
- Hand size
- Hand centering
- In-frame checks
ML Classification
Supported models:
- Random Forest
- SVM
- KNN
- MLP
Rule-Based Recognition
The rule-based engine works without a training dataset and can provide gesture predictions before a custom ML model is trained.
Hybrid Recognition
Recognition may use:
- ML model
- Rule-based model
- Hybrid agreement boost
Prediction Output
The recognition engine returns:
- Status
- Gesture
- Confidence
- Confidence level
- Alternatives
- Recognition engine
- Latency
- Advice
- Landmarks
- Hand quality
The recognition pipeline uses MediaPipe landmarks, a normalized 42-D feature representation, ML and/or rule-based prediction, confidence gating, alternatives, and live temporal smoothing.
Recognition Workflow
Webcam / Uploaded Image
↓
OpenCV Image Processing
↓
MediaPipe Hands
↓
Detect 21 Hand Landmarks
↓
Quality Analysis
↓
Normalize Hand Landmarks
↓
42-D Feature Vector
↓
ML Model / Rule-Based Engine
↓
Gesture Prediction
↓
Confidence Score
↓
Alternative Predictions
↓
Advice + Latency + Quality
↓
Optional Recognition History
The documented recognition architecture follows OpenCV processing, MediaPipe landmark detection, 42-D feature extraction, ML/rule classification, and result reporting.
ML Training Modes
Mode 1 — Pre-Trained Model Upload
- Upload
.joblibmodel - Upload
.pklmodel - Feature dimension must be 42
- Optionally retrain using existing dataset
Mode 2 — Dataset-Based Training
- Upload labeled gesture images
- Upload ZIP dataset
- Split dataset
- Select algorithm
- Train using uploaded images
Mode 3 — Default Training
- One-click model training
- Uses all valid dataset images
- Default algorithm: Random Forest
- Algorithm can be changed
- Graceful failure when data is insufficient
- Rule-based engine remains available when ML training is unavailable
These three training modes are documented in Admin → ML Model.
Technology Used
| Layer | Technology |
|---|---|
| Programming Language | Python 3.10+ |
| Backend | Django 4.2+ |
| Database | SQLite |
| Frontend | Django Templates |
| UI Framework | Bootstrap 5 |
| Image Processing | OpenCV |
| Hand Detection | MediaPipe Hands |
| Numerical Processing | NumPy |
| Machine Learning | scikit-learn |
| ML Algorithms | Random Forest, SVM, KNN, MLP |
| Model Persistence | joblib |
| Image Handling | Pillow |
| Metrics / Helpers | pandas |
| Authentication | Django Session Authentication |
| Roles | Admin, User |
| Media Storage | Local media/ |
| Model Storage | media/models/ |
The source stack is Python, Django, SQLite, OpenCV, MediaPipe Hands, NumPy, scikit-learn, joblib, Pillow, pandas and Bootstrap 5.
Most Important – Setup Instructions
1. Download the Media and Model Files
First, download the Media and Model files from the link below:
https://drive.google.com/file/d/1C45Eeh6ysT7sQ0drEaxMACrrKuqLWLes/view?usp=sharing
After downloading, extract the files if required.
You should have:
-
One Media folder
-
One
.h5model file
Place both the Media folder and the .h5 model file inside the root folder of the project.
2. Open Command Prompt / Terminal
Navigate to the project directory:
cd "C:\Project\Python Source Code\Hand Gesture Recognition System Using Python and Machine Learning"
3. Create a Virtual Environment
python -m venv venv
4. Activate the Virtual Environment
venv\Scripts\activate
5. Install Required Dependencies
pip install -r requirements.txt
6. Run Django Database Migrations
python manage.py migrate
7. Load the Initial Gesture Data
python manage.py loaddata initial_gestures
8. Create the Default Admin Account
python manage.py create_default_admin
9. Start the Django Development Server
python manage.py runserver
10. Open the Application
Once the server starts successfully, open:
http://127.0.0.1:8000/
Login Page
To access the login page, open:
http://127.0.0.1:8000/accounts/login/
Important Note
Make sure the dataset folder and .h5 model file are placed correctly in the project root directory before running the application.
The documented installation process uses:
-
Django database migrations
-
The
initial_gesturesfixture -
The
create_default_adminmanagement command -
Django's built-in development server
It is recommended to run all commands after activating the project's virtual environment (venv).
README Included
Setup Support Available
| Role | Username | Password |
|---|---|---|
| Admin | admin |
admin123 |
| User | None by default | Register from application |
The default admin is created through:
python manage.py create_default_admin
Change the default admin password before using the application in a live environment.