Shinjan Saha
This web application verifies logo authenticity using a CNN-based model, performing binary classification (real vs. fake) and multiclass classification (brand detection) with MobileNetV3 in TensorFlow. A custom dataset of over 3,000 images was collected and augmented using Google Teachable Machine and image preprocessing techniques to overcome the lack of quality public datasets. The app currently supports top global brands like Adidas, Nike, and Gucci with high accuracy, and is designed to scale to additional brands.
- Shinjan Saha
- Built and trained the MobileNetV3 CNN model for logo classification
- Created and augmented a custom dataset (3K+ images)
- Integrated the trained models into the Flask backend
- Designed a responsive web interface using HTML, CSS, and JavaScript
- Implemented real-time prediction for logo authenticity and brand detection
- 🔍 Classifies logos as real or fake
- 🏷️ Detects brand if the logo is authentic
- ⚡ High-accuracy predictions for multiple brands
- 🌐 Responsive web interface for real-time interaction
- 🛠️ Custom dataset creation and augmentation for reliable training
- Backend: Flask
- Frontend: HTML5, CSS3, JavaScript
- ML Tools: TensorFlow, Keras, NumPy, OpenCV
- Deployment Ready: Full-stack integration with Flask
bash
Copy
Edit
git clone https://github.com/Code-r4Life/Logo-authenticity-webapp.git
Navigate to the project folder:
bash
Copy
Edit
cd logo-authenticity-webapp
(Optional) Create a virtual environment
bash
Copy
Edit
python -m venv venv
source venv/bin/activate # For Linux/Mac
.\venv\Scripts\activate # For Windows
Install dependencies:
bash
Copy
Edit
pip install -r requirements.txt
Run the app:
bash
Copy
Edit
python app.pyI build smart, ML-integrated applications and responsive web platforms. Let’s build something powerful together!