🛡️ Phikita - Phishing Website Detection using ABS-CNN

A full-stack Django-based web application that detects phishing websites using a deep learning model (ABS-CNN). This project is powered by a custom-trained neural network and provides real-time prediction, visualization, and user tracking.

---

🔍 Project Overview

Phikita helps detect whether a given URL is legitimate or a phishing attempt using advanced deep learning techniques. It features:

• User Authentication System (User & Admin)
• Real-time URL prediction with confidence score
• Admin dashboard with model accuracy charts and user activity logs
• Dataset upload & retraining interface
• Contact form for user queries

---

🧠 Model: ABS-CNN (Attention-Based Convolutional Neural Network)

We developed a custom CNN architecture enhanced with attention mechanisms to give higher importance to phishing-relevant features.

Model Highlights:

• Dual Conv1D layers with BatchNormalization
• Attention mechanism for sequence feature weighting
• Achieved 96.7% accuracy on test data
• Balanced class weights to handle imbalanced data

---

📊 Admin Dashboard Features

• Model Accuracy Comparison Chart
• Prediction Trends Over Time
• Latest Uploaded Dataset Viewer
• Contacted User Queries Section
• Login & Logout Activity Tracker
• Registered Users Management

---

📁 Dataset Source

This project uses the dataset from the research paper:

“Phikita: Phishing Kit Attacks Dataset for Phishing Website Identification”
➡️ Download Paper

---

🧰 Technologies Used

Stack	Tools
Backend	Django, Python
ML Model	TensorFlow, Keras, Scikit-Learn
Frontend	HTML5, CSS3, Chart.js, Matplotlib
Deployment	GitHub
Database	SQLite3

---

🚀 How to Run the Project

# Clone the repository
git clone https://github.com/your-username/phishing-detector.git
cd phishing-detector

# Install dependencies
pip install -r requirements.txt

# Run the Django server
python manage.py runserver