High-Speed Video OCR & Entity Extraction API

Extract On-Screen Text & Entities From Video, Frame by Frame.

Developer-first API for processing video frame streams, detecting spatial bounding boxes, and extracting timecoded text at scale.

Live Frame Extraction Trace60 FPS GPU Pipeline

TEXT_ENTITY 99.4%

ORDER #8492-CONFIRMED

Timestamp: 00:04:12.8512 Bounding Boxes

JSON Payload Output

{
  "frame": 254,
  "timestamp_ms": 4250,
  "detected_text": "ORDER #8492-CONFIRMED",
  "confidence": 0.994,
  "bbox": [120, 80, 450, 140],
  "credits_charged": 1
}

Engineered for Developer Precision & Scale

Low-latency GPU execution powered by automated credit metering.

Frame-Level OCR Accuracy

Detect bounding boxes and sub-second timestamps with high-precision optical character recognition.

Idempotency & Tracing

Prevent duplicate processing charges with built-in idempotency keys and detailed job progress traces.

Metered Credit Subscriptions

Pay only for video duration and frame volume processed. Top up extra credits anytime without friction.

Get Started in Minutes

Create Frame Processing Jobbash

curl -X POST /api/v1/jobs \
  -H "Authorization: Bearer <prefix>.<secret>" \
  -H "Content-Type: application/json" \
  -d '{"input_asset_id":"123e4567-e89b-12d3-a456-426614174000","operation":"OCR"}'