---
license: mit
---

This repo contains input and output video files for a computer vision pipeline that identifies objects with a pretrained [RT-DETR model](https://huggingface.co/docs/transformers/en/model_doc/rt_detr).

The video file output and per-frame metadata (confidence scores, object counts, etc) can be uploaded to Nominal for collaborative, scalable data review.

Original video source: https://www.youtube.com/watch?v=0oucTt2OW7M&list=PPSV

Inspect this data in a [Nominal Workbook](https://app.gov.nominal.io/workbooks/ri.scout.cerulean-staging.notebook.47a6f456-dea4-4a09-9b5f-93df008af0e1?runRid=ri.scout.cerulean-staging.run.60052320-bcb3-4485-82c6-a001dbb2cdad&chartTimeStart=1321009871200.0&chartTimeEnd=1321010145800.0&chartTimeRangeType=absolute&tab=0946a1c7-ce3f-4234-ab0d-e062585f92e1)! (login required)