Purpose:

• For training computer vision models (Keras/TensorFlow) to recognize digits from digital utility meters (electricity, water, etc.).

Requirements:

• Type: 7-segment, LCD, or LED displays
• Volume: 12K–15K images (digits 0–9)

Format:

• Ideal: Pre-cropped single digits
• Ideal: Pre-cropped single digits

Examples:

• Existing dataset for mechanical counters:
  Kaggle: Tenth of Step of Meter Digits

Key Features:

• Bilingual clarity – Ensures wider reach among Russian/English contributors.
• Structured comparison – Highlights gaps vs. mechanical counter datasets.
• Visualizable – Space to embed actual meter images later.