Image Annotation
Jet Engine Blade Annotation
“It was great working with DataForce. They were very flexible in adapting to our requirements to deliver the best possible result.”
– Alexander Appel, Principal Solution Architect, zeroG
The Challenge
ZeroG is the expert center for data-driven AI and machine learning (ML) solutions in aviation. ZeroG´s solutions empower aviation organizations to gain fresh insights, make better decisions, and automate repetitive processes to realize measurable impact. One of zeroG’s areas of business is maintenance of jet engines. Manual inspection processes are labor-intensive, prone to omission errors, high-risk. ZeroG’s experts were looking for a partner to assist in training its ML model to detect defects on jet engine blades automatically, thus aiding the specialist engineers during inspection. In order to train such a model, large volumes of very accurately labeled data were needed.
• • • •The Solution• • • •
Having experimented with various internal and external labeling solutions, zeroG turned to DataForce for its high-quality output and breadth of diverse projects. Although DataForce has a proprietary annotation platform, zeroG chose to work in another third-party platform it was using at the time to continue the cadence of its ongoing project. DataForce then recruited, hired, and managed annotators working in the platform. Aside from the sourcing process, the DataForce team trained the team of annotators on zeroG’s guidelines and managed to successfully oversee the annotation of over 25,000 frames of jet engines to ensure quality.
To accomplish this, bounding boxes were drawn around various defects and assigned to one of the 11 defect categories provided. Experienced labelers were selected to do a quality assurance pass before the labeled images were provided to the zeroG experts for control. Client feedback was taken throughout the project and guidelines were adjusted accordingly so teams could be retrained where needed. This iterative and collaborative process was continued throughout the entirety of the project until all the annotated frames were delivered.
The Benefits
DataForce not only delivered on timeline and quality, but most importantly, addressed zeroG’s concerns by creating a collaborative approach to the project. The feedback was exchanged within the platform, by email, and on calls — which allowed DataForce and zeroG to resolve any pending questions and provide clarification. This approach gave zeroG a higher level of control over the project while still working with an external supplier.
“We managed to find a sweet spot of providing external support in a way that allowed the client to keep control over quality. We’re excited to be expanding our collaboration into new use cases and building a partnership!”
– Dorota Iskra, Senior Director, DataForce
DataForce has a global community of over 1,000,000 members from around the globe and linguistic experts in over 250 languages. DataForce is its own platform but can also use client or third-party tools. This way, your data is always under control.