South Korea-based AIMMO, an AI and data company, has updated its Smart Labeling and inspection technology so that it automatically labels data for autonomous driving and smart city use.
According to the company, Smart Labeling and inspection technology reduces the learning data generation process. The technology comprises a custom model, which applies auto-labeling using existing data from a client’s project, and a preset model, in which machine learning is completed in advance using data sets collected by AIMMO.
The company notes that following the update, its system can cover 2D object detection, semantic segmentation (able to recognize a collection of pixels), instance segmentation (able to recognize a segmented area for each detected object), and interactive segmentation (automatically segmenting data with just a few clicks).
Another notable update is the addition of 3D Object Detection, which supports lidar sensor data — an essential element in autonomous driving and smart city applications. The company claims its technology is capable of smart labeling while recognizing the side, back and front of data to be collected.
Along with an update to Smart Labeling, the company has launched a new curation function for refining projects. This function does not immediately label all the collected data, but allows it to be labeled for sorting. This enables the development of an autonomous driving model by sorting out only the necessary data as it is collected from a vehicle equipped with lidar and camera sensors, greatly improving the efficiency of labeling operations.
Doyle Chung, head of global sales at AIMMO, said, “With this update to AIMMO’s Smart Labeling and inspection technology, we can now provide a wider range of state-of-the-art services. Soon we will launch an additional smart labeling function with the SaaS type and data cloud linked. Through steady R&D and AIMMO’s expertise, we will build a systematic service able to respond quickly to the variety of needs in autonomous driving and smart cities.”