Ke
I am a designer crafting intuitive experiences that empower human to connect, learn, and thrive in digital spaces.
Information
Ke
I am a designer crafting intuitive interfaces that empower humans to connect, learn, and thrive in digital spaces.
Product Design
Standard Checkout
Standard Kiosk
Standard AI Labeling Tool
Standard Client Portal
Sidequest
AR & VR
Vast: Vol.2
Vast: Vol.1
IKEA Harajuku
Project Crane
Google Daydream
Experimental
A Knowldge Base
D.O.T for Onion OS
Misc 3D Projects
Composed by Jingquan
Information
Resume
Email
Instagram
Twitter
I am a designer crafting intuitive interfaces that empower humans to connect, learn, and thrive in digital spaces.
Human Interfaces / Speculative Design / Emerging Tech / AR / VR /
Product Design
Standard Checkout
Standard Kiosk
Standard AI Labeling Tool
Standard Client Portal
Sidequest
AR & VR
Vast: Vol.2
Vast: Vol.1
IKEA Harajuku
Project Crane
Google Daydream
Experimental
A Knowldge Base
D.O.T for Onion OS
Misc 3D Projects
Composed by Jingquan
Information
Resume
Standard AI Labeling Tool
The tool behind the curtain that conjures all the autonomous magic.My Role
Product Designer
Teammates
Ayal Prizant - Director of CV
Ryan Cook - PM
Timeline
2023 - now
Product Designer
Teammates
Ayal Prizant - Director of CV
Ryan Cook - PM
Timeline
2023 - now
Context
Standard AI Labeling Tool is a pioneering project that enhances the functionality of autonomous shopping technology by collecting human inputs for labeling various scenarios and shopper behaviors captured by our store cameras. These curated inputs significantly augment the learning process for our Standard AI's algorithm, making autonomous shopping a feasible reality.
The design trajectory of this project spans two strategic milestones: Improved Interaction Reivew & Tasks Review. Both of them intended to revamp the review process, dramatically increase efficiency, and substantially reduce costs.
Standard AI Labeling Tool is a pioneering project that enhances the functionality of autonomous shopping technology by collecting human inputs for labeling various scenarios and shopper behaviors captured by our store cameras. These curated inputs significantly augment the learning process for our Standard AI's algorithm, making autonomous shopping a feasible reality.
The design trajectory of this project spans two strategic milestones: Improved Interaction Reivew & Tasks Review. Both of them intended to revamp the review process, dramatically increase efficiency, and substantially reduce costs.
Improved Interaction Review
The first milestone is the Improved Interaction Review. The design involves an integration of our cutting-edge technology - "Interaction Detection". This feature smartly curates the footage, presenting only those videos to the human reviewer that contain potential shopper interactions as determined by the AI. With an impressive accuracy rate of 99%, the feature significantly simplifies the review process.
The success of implementing Interaction Detection with design is underlined by the fact that we've achieved a ~40% reduction in review time, and ~60% reduction in cost. Further, the user interface has been tailored to create an intuitive, seamless experience for the reviewers.
Tasks Review
The next design milestone is the Tasks Review, which undertakes a comprehensive redesign of the labeling workflow. The philosophy of 'divide and conquer' guides this stage, transforming a full-trip review into manageable bite-sized tasks. The redesign revolutionizes the efficiency of the system, projecting ~300% increase, and significantly growing the data size for our algorithm, enabling it to learn more and perform better.
By reshaping the labeling process, Standard AI Labeling Tool sets a new standard for AI data labeling and opens up exciting possibilities for the future of autonomous shopping.
