Request access to this annotated dataset

We review requests manually — usually within 1-2 business days.

This is a premium annotated dataset. By requesting access, you agree to credit UtilityMeters.ai as the data source in any research, product, or publication that uses it, and confirm you will not resell or redistribute the raw files.

Log in or Sign Up to review the conditions and access this dataset content.

Annotated-Mixed-Utility-Meter-And-Serial-Numbers-Dataset-Sample

DS_mixed_scor_ser_labeled_v1_2026 — Free 50-Image Preview

This is a free sample: 50 real photos, straight from the production dataset — no synthetic renders, no upscaled duplicates, no cherry-picked studio shots.

Buying image data blind is a bad deal. Download these 50 files, open them in your own pipeline, and judge for yourself: resolution, sharpness, glare handling, angle diversity, and how readable the digits and serial numbers actually are. If the sample passes QA, the full set will too — it came out of the same collection.

Dataset Summary

Real water, gas, electricity, and heat meters, photographed in the field under the conditions a production app will actually face: glass glare over analog reels, flash washout on LCD matrices, dust, scratched covers, faded stamped serials, skewed and inverted camera angles. Legacy mechanical dials and modern digital displays side by side, from over 100 European manufacturers.

Valid kits (image + YOLO + XML) — full dataset 11,589
Kits included in this sample 50
Label formats YOLO .txt (normalized) + PASCAL VOC .xml (absolute)
Taxonomy 35 classes — digits, mid-rotation states, serial decoders, global zones
Resolution Standardized to 640 px longest side (VRAM-friendly)
Structure Pre-formatted for Ultralytics and standard ML pipelines
Content 100% real photography — zero synthetic data

Dataset Structure

DS_mixed_scor_ser_labeled_v1_2026_sample/
├── Images/                 # JPG photos of utility meters
├── Labels_YOLO/            # normalized relative coordinates (.txt)
├── Labels_PASCAL_VOC/      # absolute pixel coordinates (.xml)
├── classes.txt             # 35-class master list
└── README_Documentation.txt

Data Fields — 35-class unified taxonomy

The taxonomy is the point: one unified label space for mechanical wheels and smart LCD panels, readings, and serial numbers — so one model, one inference pass, no device-routing architecture to build and maintain.

  • Digits — clean, settled digits on mechanical and digital registers.
  • Mid-rotation states — transition states for analog wheels caught between two digits.
  • Serial decoders — a separate character space for manufacturer serial numbers.
  • Global zones — region markers (register window, serial box, comma) for region-first, decode-second pipelines.

Full Dataset & Bonus Model Weights

This sample covers 50 of 11,589 total kits. The full, production-grade dataset — plus a bonus pre-trained neural network trained on these 35 unified classes (mAP50-95: 94%) — is available at:

👉 https://utilitymeters.ai/datasets-utility-meters/dataset-mix-counters

You don't start from a blank config file. You start from a working baseline and fine-tune from there.

Related Data from UtilityMeters.ai

Need volume instead of labels? UtilityMeters.ai also holds 20,000+ raw, unlabeled utility meter images, pre-sorted for mechanical and electronic displays — ideal for self-supervised pre-training or teams with their own in-house annotation pipeline.

Licensing Information

Datasets (images & annotations) include full commercial rights for training proprietary ML/AI applications. Pre-trained models are provided as a free bonus under GNU AGPL-3.0. Redistribution or reselling of raw files is prohibited.

Downloads last month
62