In This Article
- What are auto detections?
- How do Redact’s engines work?
- What do I do if I don’t need to redact one of the auto detections?
Overview
Veritone Redact is a cloud-based redaction software that harnesses the power of AI to quickly and efficiently redact sensitive information within your media. Leveraging Head Detection, Person Detection, and Object Detection engines, Redact significantly reduces the overall time and effort it takes to make your body cam footage, audio files, or surveillance footage release-ready.
What are auto-detections?
Auto-detections are the detections captured by Redact’s engines. These include heads and/or persons, laptops, license plates/vehicles, and notepads.
How do Redact’s engines work?
Redact’s engines are trained to scan your video frame by frame, looking for objects that appear to be a head, person, laptop, license plate/vehicle, and notepad.
For heads, the engine looks for a pair of shoulders with something round on top. With this said, it is not facial recognition and it does not differentiate between officers and civilians.
For Persons, the engine works similarly to our head engine, but will detect the entire person head to toe and shoulder to shoulder.
For our license detection, the engine looks for the plate itself using headlights and taillights. However, if a plate cannot be found, the engine will detect the front or rear of a vehicle.
Laptops are detected by scanning for a screen with a keyboard.
Finally, our engine has been trained to detect a hand holding various kinds of notepads and papers.
For any object not listed above that you are required to redact, such as a cellphone, you will need to apply a manual redaction or a User Defined Region. For more information, please refer to our help article: How to Add User-Defined Regions (UDRs)
What do I do if I don’t need to redact one of the auto detections?
During upload, you will select the engines you wish to run. You can select “Run Head and Object Detection” and/or “Run Person and Object Detection.”
Therefore, you have the option to detect either heads or persons (or both). However, because it is one engine that runs object detection simultaneously with person and head detection, you cannot select which specific objects are detected.
If you do not need to redact one or all of the objects our engine detects, you can simply use the filter feature on the Editor tab of the File Manager Page. For more information please visit our help article Filtering in Redact.
If you have any questions about our auto detections, contact the Veritone Support team at Support@veritone.com.
