– Case Study –
Video Annotation to Automate Global Supply Chains
Multi-sensory aerial tracking company required 98% accurate video annotation to improve the prediction of its AI engine
Challenge
A pioneering multi-sensory aerial tracking company is reimagining global supply chain operations through artificial intelligence. In order to quickly operationalize its AI platform, they required precise annotations of aerial images to help train their proprietary AI models for better prediction when tracking physical assets. A key challenge was annotating numerous assets within the video frames and with accuracy level of over 98%.
98% Accuracy
Results
SOLUTION
Innodata deployed a large team of 40-50 FTEs working 24X7 to annotate images in near-real time. The team was working on the client’s platform and had to be integrated into clients engineering teams. To achieve 98% accuracy, inter-operator confidence was measured by scoring annotated images and reviewing for finalization with a goal of reducing low confidence scores.
IMPACT
The client was able to improve the prediction of its AI engine by feeding it accurate and verified annotated images. The vigilant tracking of productivity, annotator agreement, and self-consistency metrics of humans-in-the-loop ensured high-quality annotation with proper contextualization and helped the client achieve higher ROI.
Meet an Expert
Our Team of Data Experts
A team comprised of data experts with extensive experience in developing AI-based data solutions for clients. Book a time that works for you and let us help develop a custom solution for your unique needs.
