Admin Login

SISTER

Screenshot Interrogation System for Traits and Equipment Recognition

Contribute Build Screenshots

Help us build SISTER's first neural network for label detection.

Upload your STO build screenshots to help train our CNN for label detection.

💡 You can submit multiple builds at once using the "Add Another Build" button.

Screenshot Requirements

  • Full resolution, unmodified originals
  • Clear UI elements without overlapping windows or tooltips
  • Complete build information for space and/or ground

Contact Information

Your email is required for consent confirmation and will not be shared publicly.

License Agreement

By submitting screenshots, you agree to license them under the CC BY-NC-SA 4.0 license for training our label detection CNN.

Build Screenshots

Build Platform

Build Type

Screenshots

Why We Need Your Screenshots

SISTER currently uses EasyOCR for text detection, but this creates a performance bottleneck that prevents us from offering SISTER as a web API.

The Performance Problem

EasyOCR works by recognizing text character-by-character, which is slow and requires significant processing power. For a web API to respond quickly enough for real-time use, we need a faster approach.

Our Solution: CNN-Based Label Detection

A CNN (Convolutional Neural Network) is a type of AI that excels at recognizing patterns in images. Instead of reading character-by-character like EasyOCR, we can train a CNN to recognize entire labels and text blocks at once.

This approach is much faster because:

  • Whole-label recognition - Identifies complete text regions in one pass instead of individual characters
  • Specialized training - Optimized specifically for STO's UI elements rather than general text
  • Efficient processing - Designed to run quickly without GPU acceleration
  • Reduced complexity - Eliminates EasyOCR's multi-step character analysis

What Your Screenshots Enable

  • Fast label detection without EasyOCR's character-by-character bottleneck
  • SISTER running as a responsive web API without GPU requirements
  • Real-time build analysis for web applications
  • Reliable performance across different UI scales and platforms

Account Options

How It Works

  1. Upload your build screenshots
  2. Agree to the CC BY-NC-SA 4.0 license terms
  3. Receive email confirmation
  4. Reply to confirm participation

Creative Commons License

By submitting screenshots, you agree to license them under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) license.

What This Means

  • Attribution: Your contribution will be credited
  • NonCommercial: Used only for non-commercial purposes
  • ShareAlike: Derivatives shared under same license

How We Use Your Data

  • Primary Use: Training the label detection CNN
  • Research: Computer vision research related to STO build analysis
  • Quality Assurance: Inclusion in our test suite

Data Storage & Privacy

  • Storage: Screenshots stored in public GitHub repositories
  • Email Privacy: Used only for consent confirmation, never published
  • Scope: Usage limited to STO build analysis tools and related research

Your Rights

  • Withdrawal: Request exclusion from future datasets
  • Permanence: Previously distributed copies remain under existing license
  • Irrevocability: License ensures dataset stability for researchers

Consent Process

After submitting your screenshots and agreeing to the license, you'll receive an email to verify your submission. Your data will be processed only after you confirm your consent via email reply.

SISTER is an open-source project. View on GitHub | Current release: v2025.05.31