Data entry ops · Production

How Move.ai is revolutionizing content creation with better video data

The problem

Move.ai needed to rapidly train multiple AI algorithms for markerless motion capture and depth keying, requiring labeled video data at speed; their models had to identify people, jersey numbers, and human limbs across varied distances, poses, and occlusions.

Workflow diagram · grounded in source
1
Training data need identified
trigger
“The team turned to Labelbox to help them spin up a labeling team, develop training data for their models, and iterate — all at top speed”
2
Video annotation pipeline set up
integration
“Labelbox's video annotation interface coupled with the Python SDK”
3
Boost workforce labels data
human_review
“Labelbox's dedicated workforce team, Boost, also provided a high standard of accuracy and quick responses from the account team”
4
Labeled data exported to ML frameworks
integration
“the Move.ai team exports their labeled data into Tensorflow or PyTorch frameworks to train their model”
5
Production deployment with TensorRT
output
“leverages TensorRT for production. All of this is complemented with powerful GPUs for effective AI computation”
Reported outcome

Labelbox's data engine enabled Move.ai to iterate twice as fast on their algorithms, accelerating their go-to-market strategy and product launches.

Reported metrics
Algorithm iteration speed2x as fast
Labeling results qualitymagnificent
Reported stack
LabelboxBoostPython SDKTensorflowPyTorchTensorRTNvidia RTX GPUsNvidia T4s
Source
https://labelbox.com/customers/move-ai-customer-story
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Labelbox's data engine enabled Move.ai to iterate twice as fast on their algorithms, accelerating their go-to-market strategy and product launches.

What tools did this team use?

Labelbox, Boost, Python SDK, Tensorflow, PyTorch, TensorRT, Nvidia RTX GPUs, Nvidia T4s.

What results were reported?

Algorithm iteration speed: 2x as fast; Labeling results quality: magnificent (source-reported, not independently verified).

How is this data entry ops AI workflow structured?

Training data need identified → Video annotation pipeline set up → Boost workforce labels data → Labeled data exported to ML frameworks → Production deployment with TensorRT.

Related data entry ops cases