Cale Dental Cale Dental X-ray Detector
Built by Ali Can Bakir Model ready

Automated detection on panoramic dental radiographs

Upload a panoramic X-ray and choose a model below. Two are currently deployed: a focused 4-class pathology detector and a wider 31-class segmentation model. Use either as a second-look aid for clinical review.

v1

Pathology baseline (boxes)

Highlights four common pathology findings — impacted teeth, caries, periapical lesions, and deep caries — as axis-aligned bounding boxes. Best when you want a tight, high-precision second look at the lesions a general-purpose model is most certain about.

DENTEX 2023 · ~705 images · YOLOv8s detection
v2

Wide-coverage segmentation

Outputs segmentation polygons across 31 classes — pathology, restorations, implants, orthodontic hardware, and anatomic landmarks. Broader coverage than v1 at the cost of more false positives on rare classes.

CellDetection · ~13,800 images · YOLOv8s-seg · 100 epochs · mAP50 mask 0.32

Upload

Clear, full-mouth panoramic image. PNG, JPEG, TIFF, BMP, or WebP. Up to 25 MB.

Findings

Detections above confidence 0.25. Tap a chip to show/hide a class, or click a finding on the image to toggle it.

Upload an image to see results.

What this model detects