“What do we like best about SUPA? The quality and thoughtful approach to requirements elicitation.”
Dmitri, ML Data Engineer at ZERO10
SUPA partnered with ZERO10, a tech company providing innovative AR and AI solutions designed to enhance the in-store shopping experience.
The Problem
ZERO10 needed to improve the precision of the footwear try-on models to deliver a more immersive and accurate user experience. Although the Client had previously utilised automated segmentation tools like SAM (Segment Anything Model), they found these methods lacking in the human-level accuracy required. The Client’s previous labeling partner could no longer meet the scale, speed, and accuracy benchmarks set, there was an immediate need to urgently find a new solution that could deliver the precise data segmentation essential for ZERO10’s advanced footwear try-on technology.
The Solution
ZERO10 partnered with SUPA to access SupaAnnotator - the annotation platform featuring built-in intelligent tools, collaborative QC, and feedback workflows that maximized efficiency and streamlined the labeling process; as well as SUPA’s dedicated group of experienced annotators, adept at complex segmentation tasks. This ensured both prompt delivery and adherence to strict quality standards to meet the Client's precise accuracy requirements of over 90%. SUPA also ensured rigorous training on Kaya, the distributed workforce training platform, took place throughout the project.
The Result
SUPA successfully delivered 15,000 segmentation data points within just 5 weeks, surpassing the agreed timeline of 6 to 9 weeks. Not only did SUPA deliver a rapid turnaround, we also exceeded the Client's accuracy requirement of 90%, achieving an outstanding accuracy of over 99%. This high-quality segmentation accuracy enabled the Client to provide the exceptional AR experience they envisioned for their users.
Read more about ZERO10's immersive AR technology for footwear here.
Let us walk you through the entire data labeling experience, from set up to export
Schedule a chat.png)
.png){kind=small}
.png){kind=small}