TABLE OF CONTENTS

Introduction

With the prevalence of video recordings from smart phones, dash cams, body cams, and conventional surveillance cameras, privacy protection has become a major concern, especially considering legislation such as the Freedom of Information Act.  

It is a frequent practice for organizations to redact sensitive information before releasing content outside. The content could be a paper document, audio file, or video.   

Redaction is important because it can shield an individual’s personal data, like social security numbers, names of minor children, and financial account numbers, by removing them from public legal documents to ensure privacy.  

Concept

Redaction in VIDIZMO is the act of obscuring Personally Identifiable Information, such as faces, persons, license plates, or any other sensitive information from uploaded files. VIDIZMO provides the ability to redact personally identifiable information (PII) in video, image and audio evidence files before sharing it with others. This can be done on our studio space through which you can create redacted versions of evidence files without affecting the original file.


Redaction in image and video works in the same way, you first have to analyze the file to detect objects in it. You can select from a list of objects to detect on the file. Analyzing the file can be done in following ways:

  1. At time of upload via Video Insights  
  2. On demand AI Processing
  3. Scanning the file for object from Studio Space 

For audio redaction you first have to create an audio segment for each place you need to redact. You can also use transcription to search and locate exact words said in the and then use it to create audio segment to further redact that. 


Understanding How The Feature Works

For object redaction in videos and images

You need to open the file in studio space and scan the the file for detecting objects. The AI tool will process the video and will point out the persons, faces, vehicles, license plates and weapons. For a more accurate and enhanced process of redaction, you can outline the custom objects as they are seen in a video using the drawing tool. Make sure that the portal administrator has enabled the redaction activity from the admin panel, otherwise, other you will not be able to run the studio space.  

You will have to select the duration of the video and objects you want to detect. Once the detection process is completed, the AI indexer will index the occurrence of selected objects such as faces, persons, weapons, vehicles and their license plates. For fine tuning the object positions you can edit the bounding boxes. In case there are any undetected objects you can manually create objects in the video or image.  


For audio reaction 

You need to open the video or audio file in studio space.

For audio redaction you first have to create an audio segment for each place you need to redact. There are two ways to create an audio segment;

  1. Manually: where you'll have to set start and end time of the segment 
  2. Using Transcription: where you can create a segment from the transcription


After you have created objects and audio segments in the studio space, you can select how you want the information to be redacted. For objects in video and image you select from blur, pixelate and solid color. Whereas, for audio you can select from mute and bleep. 

Manipulations on Segments 

The operations that can be performed on segments such as detected objects or audio segments include: 


Merge

Multiple detections of the same instance can be merged. This feature is helpful if the automatic detection has labeled different occurrences of same instance as different objects.

 

Note: Detected objects can be merged in video only.



Delete

An occurrence that has been selected automatically which does not need to be redacted can be deleted. 


Edit

The AI tool detects and names the detected instances as person 1, object 1 etc. These names can be edited to identify exactly what detected instances are being redacted.  


Object Detection Classes 

Automatic Face Redaction 

The AI video indexer begins to search and detect faces whenever a new redaction scanning activity is initiated by the user.  

When the process is complete, a thumbnail image of each detected face is then displayed, these thumbnails can be used to select the individuals that should be redacted from the scene. 

The processing time for auto face detection may vary depending on the video file size and will be affected by the resolution, length, frame rate, and other factors related to video. The success of the auto face detection can vary depending on the quality of the video and whether the subject is facing to the front or the side.  


Automatic Person Redaction  

Like face detection, while the video is being processed by the indexer for detection, it will detect persons appearing in the video frames and will show a thumbnail in the persons tab. Merge, edit, and delete actions are available for detected persons to enhance the redaction activity.  


License Plate Redaction 

The automatic detection tool can detect license plates in the video frames. License plates are also a part of PII so this sensitive information can be automatically detected and redacted from a digital evidence file.

 

Manual Object Redaction  

To make the process more efficient, users can mark custom objects in the video using the drawing tool to have them redacted. For example, there is an appearance of a knife in the video and you want to hide it from viewing, you can simply outline the knife by using the drawing tool. Your outlined object will appear in the objects tab of the redaction studio and you will be able to blur, pixelate or put a solid box on the selected object. 


Redaction Studio Layout and Controls

Following is the screenshot of the studio space. These are the tabs where detected instances are shown to the user. Here, merge, edit, delete, and view options are available as discussed earlier in this article.

 


 


Use Cases

So far, the feature is discussed and a good understanding has been developed. Now, let us understand where this whole redaction process will be helpful. What can be achieved using these redaction tools in your video, how can you group them together for protecting sensitive information.   


Redaction for Law Enforcement

Handling digital evidence is so important for better law enforcement. Hence, Law enforcement officers needs to redact and enhance all types of video footage, audio files, images, and PDFs from any source, including body cameras, in-car cameras, cellphones, and more. Designed to handle all proprietary formats and media, VIDIZMO makes it easy to redact sensitive documents and recordings, no matter what the format of the file is. 


Redaction for Education

Education providers needs to comply with the Family Educational Rights and Privacy Act (FERPA). Protect personally identifiable information (PII) by removing identifying data. Education providers can easily redact students faces, people (i.e., entire bodies) and identifying characteristics like badges in streamed classroom sessions or any video footage.  


Redaction for Healthcare

Protecting your PHI (Protected Health Information) does not have to be hard. Whether you need to redact medical records or de-identify PHI from a research study or clinical trial, VIDIZMO’s powerful and easy to use redaction tool can help you. 


Redaction for Government

Governing is hard enough; redaction does not have to be. Our strong security means your data stays safe, and a comprehensive audit log is secured throughout the editing process meaning agencies can focus on mission-critical tasks instead of managing unnecessarily complicated redaction workflows. Compliance officers, FOIA specialists, and records specialists can automatically redact complex video footage with thousands of faces, license plates, papers, vehicles, bodies, and screens. 



Read Next