AI-based Video De-identification Masking Workspace Redesign
Redesigning frame-by-frame mask review with clear tools, contextual actions, and faster navigation.
AI Leaders
2024
✨Summary
PROBLEM
The original workspace grouped actions without regard to context and
required users to navigate frames one by one, making the work tedious.
UX GOAL
Help non-technical users quickly review and edit AI-generated masks through self-explanatory, low-friction interactions.
SOLUTION
Redesign the workspace interaction based on the usage context and the menu usage frequency.
Preview toggle / Immediate Action Buttons / Frame Jumps / Keyboard Shortcuts.
TESTING
In brief testing with 5 users unfamiliar with the service to test the toolbar design variations. the labeled toolbar was easiest to identify and use.
Background & Goal
Streamlining frame-by-frame review
The AI-based video de-identification desktop software automatically detects and masks people in footage, built to meet South Korea's data privacy guidelines.
The AI's results still needed a human review and edit — however the process had frictions that slowed down the work. Thus, the project aimed to redesign the interaction that can make the work frictionless and efficient.
Non-technical school staff
Not familiar with video editing, masking tools
2 / AI detection
Flags objects to mask
3/ Frame review
User edits masks
4/ Export
Export footage
Interaction Spot
1 / Upload video
Footage uploaded
User Needs
Clear action / Feedback
Efficient, frictionless review
Problem
The original workspace grouped actions without regard to context and required users to navigate frames one by one, making the work tedious.
The original interface had several friction points that confused users and slowed the review process.
Workspace actions were grouped into one panel regardless of when users needed them.
This confused first-time users and made relevant actions harder to identify.Users also had to click through frames one by one, adding repetitive work to an already detailed review task.
Design Solutions
Organizing tools around context and frequency
The workspace toolbar and major actions were redesigned around usage frequency and context.
Design Exploration
Self-explanatory toolbar panel for the first time users
Initially, we explored toolbar designs that can provide more space and visual cleanness for the workspace.
Three toolbar variants were tested with 5 users unfamiliar with the service.
-> Unlike the initial assumption, the toolbal with labels was easier for users to identify and use without frictions.
