Datasets:

bandicootlab
/

smartphone-testing-data

license: other
language:
  - en
tags:
  - smartphones
  - mobile-devices
  - hardware-benchmarks
  - camera-testing
  - display-testing
  - battery-life
pretty_name: Bandicoot Lab — Smartphone Lab Test Data (Sample)
size_categories:
  - n<1K
configs:
  - config_name: devices
    data_files: devices.csv
  - config_name: scores
    data_files: device_scores.csv
  - config_name: test_runs
    data_files: test_runs.jsonl

Bandicoot Lab — Smartphone Lab Test Data (Sample)

Lab-measured smartphone test data: raw measurements and computed 0–950 performance scores across display, camera, battery, performance, charging, audio, biometrics, and connectivity.

This sample contains the complete dataset for 3 of 51 devices spanning 10 brands in the full catalog. The three devices were chosen to span the market:

Samsung Galaxy S25 Ultra — $1,299 flagship (200MP main, dual telephoto)
Apple iPhone 17 — $799 mainstream flagship
Motorola Moto G Power (2026) — $300 budget device Each device is measured the same way the full catalog is: per-lens camera sharpness, color, and dynamic range across three lighting conditions; per-patch color accuracy on a 24-patch chart; per-step dynamic range; full per-percent charging curves; speaker frequency-response and distortion sweeps; display brightness, color, and tone-mapping; and more. The three sample devices carry every metric and sub-score the full catalog does — nothing is stripped.

Files

devices (devices.csv) — device specifications, one row per device.
scores (device_scores.csv) — every computed score and sub-score, including per-lens camera scores for devices with multiple cameras.
test_runs (test_runs.jsonl) — raw per-test measurements with full nested metric detail (per-patch, per-step, per-percent). One row per test run.

Scoring

Scores are computed hierarchically: raw metrics roll up into sub-scores, which roll up into category scores, which combine into an overall device score. This sample publishes the resulting scores and the measurements behind them. The scoring formulas are proprietary and are not included.

Full catalog & licensing

This is an evaluation sample. The full catalog of 51 devices — and commercial licensing for OEMs, analysts, retailers, and AI/ML teams — is available at https://bandicootlab.com.

Contact: christian@bandicootlab.com

License

Evaluation and research use only. No commercial use. No redistribution. Commercial licensing is available — see the link above. Full terms in LICENSE.