Data entry ops · Production

Advertima accelerates video data labeling 10-15x with super.AI

The problem

Advertima's cashierless checkout solution required large-scale video labeling, but their open-source tool CVAT was slow, buggy, and had long turnaround times, generating significant overhead when ingesting proprietary camera data.

First attempt

CVAT, the open-source video labeling tool Advertima was using, proved too slow and buggy with limited UX and long processing turnarounds, making it unfit for their proprietary camera footage volume.

Workflow diagram · grounded in source
1
Video footage submitted via API
trigger
“We plugged in the customer's video footage to our API”
2
AI labels video content
ai_action
“to label things like people, products and people interacting with products”
3
Processed labeled data returned
output
“process hundreds of hours of video footage for them in an easier and faster way”
Reported outcome

super.AI processed hundreds of hours of video footage 10-15 times faster than CVAT with greater accuracy.

Reported metrics
Video processing time10-15 times
Labeling accuracygreater accuracy
Reported stack
super.AICVAT
Source
https://super.ai/case-studies/digital-signage-company-accelerates-video-labeling-with-super-ai
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

super.AI processed hundreds of hours of video footage 10-15 times faster than CVAT with greater accuracy.

What tools did this team use?

super.AI, CVAT.

What results were reported?

Video processing time: 10-15 times; Labeling accuracy: greater accuracy (source-reported, not independently verified).

What failed first in this deployment?

CVAT, the open-source video labeling tool Advertima was using, proved too slow and buggy with limited UX and long processing turnarounds, making it unfit for their proprietary camera footage volume.

How is this data entry ops AI workflow structured?

Video footage submitted via API → AI labels video content → Processed labeled data returned.