Case Study: C-SPAN achieves fast, accurate facial-recognition indexing of video archives with Amazon Web Services (Amazon Rekognition)

C-SPAN - Customer Case Study

C-SPAN, the nonprofit broadcaster that archives and streams U.S. government proceedings, faced a labor-intensive challenge: manually identifying speakers in its video archive. A small team of two to three indexers could only tag about half the footage at a rate of roughly one hour of labor per hour of video, limiting searchability for journalists, researchers, and the public.

C-SPAN implemented Amazon Rekognition in a simple S3 → SQS pipeline that uploads six‑second screenshots and matches them against a 97,000-person image collection (indexed in under two hours), using a 96% confidence threshold and shot detection to cut duplicates. The automated system delivers human-level accuracy at least twice as fast—reducing indexing time from one hour of work per hour of video to about 20 minutes—and will enable 100% indexing of live content and large-scale archive processing.

C-SPAN

Alan Cloutier

Technical Manager