Innodata Expands AI Capabilities to Support Robotics Initiatives

Dexai, a leader in commercial kitchen robotics, engaged Innodata to create training data that will allow its flagship robot, Alfred, to detect ingredients, prepare meals, and recognize spills to prevent food waste. Rollout into commercial kitchens will take place in 2021, and each new installation will require training on new parameters – ingredients, kitchen layout, and types of spillage. Training requires semantic segmentation of thousands of images across dozens of ingredient types in turnaround times as short as two weeks. Innodata plans to quickly scale this effort while guaranteeing 95% pixel accuracy. According to Dexai data scientist Alan Zhou, “Combining high quality data from Innodata with Dexai’s state of the art machine learning algorithms will help empower Alfred’s computer vision models to learn faster with increased accuracy.”