Quality assurance · Production

Sharper Shape builds streamlined annotation pipeline with Labelbox to detect utility defects

The problem

Sharper Shape relied on heavily manual workflows and open-source labeling tools that lacked the required configuration, customer support, and infrastructure management capabilities needed to meet their SLAs and scale training data production.

First attempt

Open-source labeling tools did not provide sufficient configuration or customer support for Sharper Shape's annotation needs, and manual experiment tracking added overhead.

Workflow diagram · grounded in source
1
Raw data ingested via API
integration
“connect their raw data into Labelbox via a simple API”
2
Internal and external labelers annotate
human_review
“Labelbox's collaboration features also enabled rapid onboarding, training, and throughput for both internal and skilled external labelers to work together in one centralized environment”
3
Model-assisted label generation
ai_action
“model-assisted labeling, which allows teams to import their model into Labelbox and address edge cases”
4
Labeler reviews false positives
validation
“cutting the labeling load of our labelers to that of reviewing for false positives”
5
Labels fed back into Labelbox
feedback_loop
“run a loop to generate labels from our model's inference, and feed those back into Labelbox”
6
Model building and deployment
output
“concentrate on model building and deployment, without sparing additional engineering effort”
Reported outcome

Sharper Shape cut labeling costs by as much as 50%, sped up model training by over 10X, and can now concentrate on model building and deployment without additional engineering effort.

Reported metrics
Labeling cost reduction50%
Model training speed improvementover 10X
Training data qualitymaintaining the highest quality
Expected time and cost reduction from model-assisted labelinganother huge reduction in time and costs
Show all 5 reported metrics
labeling cost reduction50%
model training speed improvementover 10X
training data qualitymaintaining the highest quality
expected time and cost reduction from model-assisted labelinganother huge reduction in time and costs
model accuracy improvementincrease our capabilities and model accuracies exponentially
Reported stack
Labelbox
Source
https://labelbox.com/customers/sharper-shape-customer-story
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Sharper Shape cut labeling costs by as much as 50%, sped up model training by over 10X, and can now concentrate on model building and deployment without additional engineering effort.

What tools did this team use?

Labelbox.

What results were reported?

Labeling cost reduction: 50%; Model training speed improvement: over 10X; Training data quality: maintaining the highest quality; Expected time and cost reduction from model-assisted labeling: another huge reduction in time and costs (source-reported, not independently verified).

What failed first in this deployment?

Open-source labeling tools did not provide sufficient configuration or customer support for Sharper Shape's annotation needs, and manual experiment tracking added overhead.

How is this quality assurance AI workflow structured?

Raw data ingested via API → Internal and external labelers annotate → Model-assisted label generation → Labeler reviews false positives → Labels fed back into Labelbox → Model building and deployment.

Related quality assurance cases