Railway Track Defect Detection System

Dual Cross-Verifying Geo-Marking System for railway track defect detection. This advanced system integrates both microphone-based and ultrasonic transmitter-receiver setups to provide robust and accurate monitoring. The data is used by the AI model that is integrated to the Cloud server and thereby automatically detects and geo-marks the defects detected. The defect data is encrypted for better security and can only be accessed by select individuals.

Key components, system highlights, team details, project Links, instructions on how to run the project locally and tech stacks & technologies used have all been mentioned below.

---

Key Components

1. Detection & Data Collection

• Audio Signals: Captures and clips audio into 5-second files.
• GPS Coordinates: Associates GPS data with audio files.
• Data Transmission: Sends data to the cloud using an IoT device.

2. Data Storage

• Base64 Encoding: Encodes audio files and stores them securely in DynamoDB.

3. AI Processing

• Data Retrieval: Retrieves data from DynamoDB to AWS Elastic Beanstalk.
• Feature Extraction: Extracts MFCC features.
• Algorithm: Analyzes data using a Logistic Regression and Random Forest pipeline.

4. Anomaly Detection

• Defects Detected: Identifies wear, cracks, and burnt wheels.
• Geo-tagging: Marks anomalies with GPS coordinates and stores them in a secondary database.

5. Visualization

• Dynamic Map: Displays geo-tagged anomalies in real-time on an integrated geomap.

---

System Highlights

• Cross-Verification: Enhances accuracy with combined microphone and ultrasonic data.
• Data Transmission: Ensures redundancy with GPS modules, Raspberry Pi, and cloud servers.
• Automatic Detection: Uses AI models to detect and geo-tag defects.
• Secure Access: Provides encrypted data for secure access by maintenance crews.

---

Project Links

• Early Stage Presentation: View the PPT
• Video Demonstration: SIH_1584_demo_video
• 3D CAD Models: Acoustic Mic System
  Ultrasonic Transmitter-Receiver System
• Source Code: GitHub Repository
• Additional Resources: Research Paper

---

Running the Project Locally

1. Clone the Repository

   git clone https://github.com/mohammadBilal03/SIH_INTERNAL_ROUND_2_TopCoder.git
   cd SIH_INTERNAL_ROUND_2_TopCoder
   

2. Install Dependencies

   npm install
   

3. RUN THE PROJECT

   npm run dev
   npm run start
   

Backend: Run npm run start to host the backend locally at port 5000.

Frontend: Run npm run dev to host the frontend locally at port 5173.

---

Tech Stacks and Technologies used:

SOFTWARE:

• Application Development: React Native Framework with HTML, CSS & JavaScript Android Studio for Emulator

• Website Development: Vite JS and React framework (frontend). Express (Node.js web application framework) (backend).

• Machine Learning Frameworks: Logistic regression Librosa library Random forest Decision tree AWS Elastic Beanstalk Flask (for deployment)

• Cloud Services: AWS IOT Core AWS Lambda AWS SNS AWS Dynamo DB AWS Cloud services

HARDWARE:

Raspberry Pi 5 INMP 441 NEO 6M module (GPS) Ultrasonic Transmitter & Receiver NC Mic (Shure SM7B) Defect Detecting mic (Sennheiser MKE 600)