Linker Networks Inc. announced the launch of its service to deliver AI-based auto-labeling with continuous learning for the autonomous driving industry. Linker's auto-labeling technology integrated with Microsoft Azure cloud services is designed to deliver a seamless and scalable solution to pre-process datasets used for machine learning in vertically integrated industries such as autonomous driving, manufacturing, public safety, and medical imaging. The service will be generally available by June 2019.
The company designed the system with the understanding that the pre-processing of datasets for AI machine learning is currently labor intensive, time consuming, and highly customized to individual labeling requirements. To train real-time inference models for autonomous driving today, customers manually label their data sets, which can be expensive as well as difficult to replicate for new data and object recognition scenarios.
Linker develops AI models with continuous learning to improve auto-labeling model accuracy and scalability. Its auto-labeling service uses a human workforce to deliver ground truth for images and videos with speed and automation. By integrating Microsoft Azure Machine Learning service, Linker says it can improve productivity and reduce costs with autoscaling compute, tune hyperparameters faster, and deploy customer-owned auto-labeling AI models tailored to specific requirements efficiently.
Autonomous driving platforms generate hundreds of millions of images and videos to be labeled with the highest levels of accuracy. Linker says that reducing labor cost and time-to-label is critical for these platforms' competitive market advantage.