LENS
▸ see what the machine sees
server · onnx + tesseract · $0

LIGHT TABLE

Drop an image on the table — or paste / pick one.

READOUT

CLASSIFY

READ

PALETTE

MODEL—
CLASSIFY—
OCR—
DIMENSIONS—
CONFIDENCE—
WORDS—
server · onnx + tesseract · no Vision API · $0

Every analysis runs inside this Vercel function: a real ONNX vision transformer names the image, Tesseract reads its text, and the palette is quantized straight from the pixels — no Google/AWS Vision API, no API key, $0.

Processed in-memory on our server. Nothing is stored.