Skip to content

Readme.md

Latest commit

 

History

History
113 lines (81 loc) · 3.81 KB

Readme.md

File metadata and controls

113 lines (81 loc) · 3.81 KB
Raw

AstraSafe: AI-Powered Space Station Safety Object Detection

🛰️ Event -
👨‍🚀 Team Name - CODE2AIM 💡 Problem Statement - Safety Object Detection in Confined Space Environments
📩 Team Leader Email -

🌠 A Brief of the Prototype
image

AstraSafe is an AI-powered computer vision system designed to detect, monitor, and assess safety-critical objects inside space stations.

It uses a custom-trained YOLOv8 model to automatically identify key safety objects, display confidence scores, and draw bounding boxes over detected objects.

The platform supports both live camera feed detection and image uploads, making it ideal for real-time monitoring as well as offline audits.

🌍 Modules Overview image

🧠 1. Live Camera Feed Detection

  • Uses device camera for real-time inference
  • Captures frames every second and predicts safety objects
  • Draws bounding boxes with object names and confidence scores
  • Dynamically updates the Recent Detections panel

💡 Simulates continuous safety monitoring for astronauts in mission-critical environments.

☁️ 2. Image Upload Detection image

  • Click or drag-and-drop to upload images
  • Runs YOLOv8 predictions on uploaded images
  • Draws bounding boxes on detected objects
  • Displays confidence scores for each detection

📸 Useful for offline inspections and safety audits.

📊 3. Detection Stats

  • Shows how well the model performs during training
  • Key metrics from training:
Metric Value Meaning
Precision 0.841 Accurately identifies safety objects without false positives
Recall 0.682 Finds most true positives in frames
mAP@0.5 0.742 Detection accuracy at 50% IoU
mAP@0.5–0.95 0.596 Robustness across stricter IoU thresholds
Fitness 0.596 Overall balanced performance

🧠 Interpretation:

  • Confidently identifies most safety-critical objects (OxygenTank, Extinguisher, etc.)
  • Good generalization with minimal overfitting
  • Ready for inference & demo

🧩 Safety Objects Detected

  • OxygenTank
  • NitrogenTank
  • FirstAidBox
  • FireAlarm
  • SafetySwitchPanel
  • EmergencyPhone
  • FireExtinguisher

✅ Model achieves strong and competitive detection metrics for all 7 classes.

🧰 Tech Stack

Layer Technologies
Frontend HTML · CSS · JavaScript · Leaflet.js
Backend Flask
AI & ML YOLOv8 · Ultralytics · NumPy · PIL

👨‍👩‍👧‍👦 Meet the Team – CODE2AIM

Member Role
Pratham Ranjan AI/ML Engineer / Backend
Ishani Jindal Frontend Developer/ UI UX Designer
Aditi Mehta Frontend Developer/ UI UX Designer

✨ Together, Team CODE2AIM ensures astronaut safety through AI-powered monitoring systems.

🧩 Code Execution Instructions

▶️ How to Run the Frontend

# 1. Clone the Repository
git clone https://github.com/prathamranjan05/Safety-Object-Detection.git
cd Safety-Object-Detection

# 2. Open index.html in browser
# Or serve via simple HTTP server
python -m http.server 8080

SCREENSHOTS

Features: -

image

Uses: -

image