LOCAL RESEARCH · NO PHOTO UPLOAD
MedCam Photo Validation Lab
Compare the current experimental MedCam image-quality rules against independent human review of your own permission-cleared medical-film photographs. This is photography-quality research, not a medical diagnosis or clinical certification.
Privacy: selected photos remain in this browser session. No server upload or medical AI API. Exported files contain anonymous case IDs and reviewer labels—not photo bytes, actual filenames, EXIF, names or GPS. Use patient photographs only with appropriate permission.
1. Choose photos and context
JPG, PNG, WebP • up to 25 MB / 48 MP each. Larger files are skipped.Choose local photos to begin.
Reviewer should check whole-film visibility, labels, focus and reflections. An acceptable photo is not certified diagnostic imagery.
2. Review each photo
No photo loaded.
3. Anonymous quality statistics
The static-frame candidate rule deliberately excludes motion. These disagreement rates do not measure actual live Auto Capture accuracy or clinical readability.
4. Real-world validation protocol
Collect permission-cleared photos across real film types, low/high-end cameras, lighting and devices. Record independent human review outcomes. Log whether actual live Auto Capture fired, whether a retake was needed, and the browser and OS version.
Keep improvement and final test datasets independent. Avoid repeated photos of the same study across training and held-out test partitions. Never automatically train on private patient photos.