🤖 Team Ray – QR Code Detection & Analysis Module

Welcome to the QR Code Detection & Image Processing Module for Team Ray, the University of Huddersfield’s aerospace team competing in the UKSEDS Olympus Rover Trials (ORT) 2024–25.

This codebase powers the AI-driven vision system for our Mars Rover prototype. It simulates how the rover detects, decodes, and prioritizes QR codes placed on Martian rocks at geological points of interest (POIs).

📌 Built & Maintained by:
🧑‍💻 Anzer Khan – Software Lead, Team Ray
🎓 MSc Artificial Intelligence, University of Huddersfield

---

🧠 Project Purpose

This module enables our rover to:

• 📸 Capture still images of Martian terrain (rocks with QR codes)
• 🧠 Detect multiple QR codes per image
• 🔍 Decode all QR codes and select the smallest valid one
• 📦 Generate a structured summary packet for submission

It is a mission-critical component of the imaging pipeline, directly influencing scoring in the competition.

---

🌌 Bigger Mission Context: Olympus Rover Trials 2024–25

The UKSEDS ORT simulates a Mars pre-landing reconnaissance mission. The rover must:

• Navigate rugged Martian terrain 🌄
• Detect rocks with QR codes 🪨📷
• Send back images and decoded data via wireless comms 🌐

🔬 Scoring is highest when we:

• Capture the smallest successfully scanned QR 📈
• Submit both the image and decoded content 📑

This repository simulates the AI-powered image processing part of that workflow.

---

🧪 Live Testing with Streamlit

You can try out the interactive web app here:

👉 Live Streamlit App – Try It Now

To run locally:

streamlit run app.py

🛠️ What’s in This Repo?

Module/File	Description
image_processing/	Core logic: QR detection, cropping, decoding, summary generation
qr_dataset/	QR images used for generating test scenarios
qr_code_images/	Output folder for generated composite test images
cropped_qr_codes/	Saved cropped versions of detected QR codes
app.py	🧪 Streamlit web app for testing the pipeline interactively
main_pipeline.py	CLI-based execution of the core pipeline
config.py	Configurable parameters (QR sizes, paths, stride, margin, etc.)
Dockerfile	Container config for reproducible deployment
requirements.txt	Python dependencies

---

🚀 How the Pipeline Works

1. Image Input: Accepts either:

   • Generated synthetic image with multiple QR codes
   • Uploaded real-world test image (simulating camera capture)

2. Detection: Detects all visible QR codes using OpenCV and Pyzbar

3. Decoding & Ranking:

   • Decodes each QR code
   • Calculates bounding box area
   • Chooses the smallest successfully decoded QR

4. Summary Packet:

   • Stores QR data and metadata in a .json file
   • Saves a bounding-box annotated image
   • Optionally saves cropped version of QR

5. Visualization:

   • Streamlit app allows team to test, visualize, and download results live

---

✅ Features:

• Generate and download random QR images
• Upload real rover images for detection
• Annotate and view all detected QR codes
• Automatically highlight the best-scoring QR
• View or download the .json summary

---

📦 JSON Output Example

{
  "timestamp": "2025-05-04T15:23:00",
  "image": "temp_upload.png",
  "total_detected": 3,
  "decoded_qrs": [
    {"content": "SiO2-78%", "bbox": [100,120,80,80], "area": 6400},
    {"content": "Fe2O3-12%", "bbox": [230,180,60,60], "area": 3600}
  ],
  "best_qr": {
    "content": "Fe2O3-12%",
    "bbox": [230,180,60,60],
    "area": 3600
  }
}

---

🧩 How This Fits Into the Full Rover Pipeline

Stage	System	Description
Image Capture	📷 Digital Camera (OV5640)	Triggered via ESP32, takes still shot of POI rock
Image Transmission	📡 Wi-Fi Module	Sends image to base station
Image Processing (this)	💻 Python + OpenCV + Pyzbar	Detects, decodes, and summarizes QR code data
Command Submission	🧠 Manual or automated upload	Image + data sent to judges

---

🛠️ What I'm Currently Working On

• ✅ Sliding window multi-QR detection

• ✅ Smallest QR selection based on bounding box area

• ✅ Streamlit app for testing

• 🔄 Next Up:

  • JSON submission generation
  • Camera integration simulation (OV5640 → pipeline hook)
  • Resilience testing on noisy/blurred images

---

👨‍💻 Skills Demonstrated

• 🧠 Computer Vision – OpenCV pipelines, multi-object detection
• 🤖 AI Readiness – Laying groundwork for ML-based quality scoring
• 🌐 Software Engineering – Modular code, config-driven design
• ⚙️ Tooling – Docker, Streamlit, real-time testing UX
• 📈 Team Collaboration – Used by teammates to validate detection success

---

📌 Future Improvements

• Real-time camera feed integration via socket or FTP
• Confidence scoring for ambiguous QR scans
• Integration with mission control dashboard

---

👋 Get in Touch

📧 anzerkhan27@gmail.com 🔗 LinkedIn – Anzer Khan 🏛️ University of Huddersfield – MSc Artificial Intelligence

---

This repository is proudly maintained as part of the Mars Rover mission by Team Ray. Built with clarity, purpose, and space-grade ambition. 🚀