Run trained vision models on local silicon.

Choose from 44+ specialized detection models compiled into NCNN format for edge inference. Fast, zero-overhead execution with single-operator exclusivity.

Active Selection NCNN FP16

Sahil Kapase YOLOv8 Pothole Detector

Vision Catalog

44 Models Cataloged

Visual Viewport

Loading Model...

NCNN FP16

0.0 FPS

0.0 ms

CONFIDENCE: 35%

Active Localizations 0 Localized

Awaiting video frame input...