My contribution and takeaways

• computer vision

• interaction design

• user interviews

• competitor analysis

Problem

Body worn cameras and CCTV (Close Circuit Television) cameras used by officers can capture all kinds of people, objects and activities in the course of an officer’s duties and responses to calls. The public or the media may make FOIA (Freedom Of Information Act) requests for some of this video footage. The police must blur out images of children, victims, personal information like license plates, store signs and other sensitive images in the video in order to protect citizens before releasing the video to the media. 

1. Manually performing such a task is labor intensive

With 1 camera for every 11 people, considering the population of Washington, D.C., 5.9 billion hours of video footage are generated every week. 

2. Video editing software requires countless hours to master, making police dependent on special video editing agencies. 

3. Editing large number of videos frame by frame is expensive

Users

 User flow

Summary of competitor analysis

Understanding user behavior from interviews

Emerging trends from user interviews

1. Speed is a crucial factor to consider when dealing with a large number of FOIA requests.

2. User must have the flexibility to add masks for redaction manually. 

4. Provide a way to preserve original video

5. Organize redacted videos by meaningful names

Exploring solutions for user needs

High fidelity prototype

Design decisions

Progress indicator within original video instead of explicit indicator while analysis / redaction is taking place

Having limited space within the editing panel, it was necessary to have as compact UI elements as possible. This style of progress indicator allowed user to view progress without moving focus from the original video. 

Enter mask creation mode and exit with a single click vs. Mask creation within video frame

Manual masks can be created by clicking on the preview panel as well as the video frame. Having this flexibility was important to some users because viewing masks filtered in a panel was easier for them to add to the list while other users felt that being able to draw on the video panel provided them with some context around the mask.

System status visible outside the application in notification panel instead of pop-up messages

Pop up messages would not provide user real time feedback of the progress of the redaction. Having a dedicated space for progress indicators would provide users real time status messages of the task taking place in the background.

Adding/deleting manual masks within video frame vs. editing panel

Giving user option to add/ delete masks on the video panel would add context to the mask being created (showing surrounding area) while showing the mask in a thumbnail view on the preview panel would show the content of the mask without noise. Since both scenarios had equal importance, it was decided that we have both options.

Additionally, hover over feedback to the user is provided by changing color of mask when user hovers over masks. User may choose to unselect mask by simply clicking on the mask.

 Single button for all operations vs having multiple buttons

Changing state of a single button for next steps would make the entire redaction process seem like a linear sequence of steps rather than disjointed pieces of events happening in the background.

Having multiple buttons on the other hand, would require disabling certain actions while a button is active. 

Outcome

maintained versions of redacted and original videos that makes it easy to locate, download and share