(주)인지

AI Vision Solution

Service Introduction

Building upon our proprietary AI-vision technology and expertise in 3D vision data processing, we deliver solutions capable of high-speed, high-precision image segmentation and 6D pose estimation, even in complex environments. Through these core technologies, we provide advanced services such as Bin Picking & Packing, logistics automation, 3D environment modeling, and SLAM (Simultaneous Localization and Mapping) for both industrial and general applications. With reliable and precise AI-vision solutions, we are committed to enhancing productivity and driving innovation in automation across diverse industries.

Key Features

• Robust image segmentation and 6D pose estimation, resilient to occlusion, stacking, and lighting variations.

• High-speed, high-accuracy simultaneous processing of 2D and 3D data, powered by our patented deep learning network technology.

• Flexible AI-vision solutions that can be adapted to a wide range of industrial and general environments.

Industrial Environment

Bin Picking & Packing

For various industrial parts randomly placed inside a box, our system performs 2D image-based object recognition and segmentation. Each identified object is then matched with its corresponding 3D point cloud to estimate its precise 6D pose (position and orientation). Based on the estimated 6D pose information, workspace modeling is performed, and the resulting data are directly utilized for robot-based automation tasks.

Conveyor Belt Environment

In industrial settings, our system performs real-time recognition and segmentation of metallic components moving along conveyor belts. Integrated with our High-Precision 3D Camera System, the solution intelligently detects each object’s regions of interest (RoI) and relays this information to the 3D camera, enabling even more accurate 3D measurement. Furthermore, by analyzing the segmented and measured 3D data, the system can conduct defect detection and quality inspection with high precision.

Logistic Environment

Household and Logistics Object Modeling

Our system performs real-time recognition and 6D pose estimation of household and logistics objects that are randomly placed on a workspace or worktable. Even in complex scenes with occlusions and overlapping objects, our proprietary 3D AI Vision Solution effectively overcomes these challenges while maintaining both accuracy and real-time performance. This technology serves as an optimized solution for logistics automation and intelligent robotic vision systems, ensuring reliable performance in dynamic environments.

General Environment

Indoor Environment Modeling

For typical indoor spaces, our system performs real-time 3D modeling and SLAM (Simultaneous Localization and Mapping). By integrating 2D object recognition with 3D object and scene modeling, the solution enables precise spatial reconstruction of indoor environments. Through the AI Vision Solution, unnecessary elements within 3D point cloud data used for SLAM — such as moving people or transparent glass doors — are automatically detected and removed, allowing for more accurate and efficient navigation map generation. Furthermore, by leveraging the deep learning–based 3D point cloud processing technology of our 3D AI Vision Solution, noise and distortion are minimized, resulting in a refined, high-precision 3D model suitable for various applications.