Notifications
Back
Vision AI for
Automotive

  Deep learning-based vision technology in the inspection sector for autonomous driving, vehicle manufacturing, and finishing process is garnering significant attention as a core technology in the industry, rapidly advancing in both scope and sophistication.

Use Cases

Vehicle Gap and flush Inspection, Data Processing for autonomous Parking & Driving, and Defect Inspection in automotive parts

More
Issues and
Needs to Address
  • The technology required to automatically process the multitude of data collected from cameras and sensors installed in vehicles for model training is necessary.
  • Technology capable of making precise assessments regarding whether each part of a vehicle, composed of numerous components, contains defects and whether they are assembled in the correct position is necessary.
  • Traditional 3D sensors and standard cameras struggle with the light refraction and transmission properties of transparent objects, leading to inaccurate measurements and undetected defects.
  • 3D technology for gap and flush measurement requires close-range imaging within specific standards due to significant constraints in the quality of 3D reconstruction.
Solutions and
Applied Technologies
  • Applied various types of road line data classification models, automatic optimization technology for Regions of Interest (ROI) within images, image noise removal technology, and Line Segmentation models.
  • Applied an anomaly detection model to confirm anomalies in parts and utilized a detection and segmentation model (SINNet) to identify these anomalies.
  • AI algorithms embedded in SAVIM, the SNUAILAB Vision Inspection Machine, effectively remove various noise data generated during 3D reconstruction and restore clear 3D images of vehicles. 
Results of Technology Implementation

01

SAVIM, the SNUAILAB Vision Inspection Machine, uses patented algorithms to minimize the overall costs of the gap and flush inspection process and supports full automation, regardless of the type and color of vehicles, under existing infrastructure conditions, including cameras and conveyor belts. 

02


Developed a model capable of identifying whether each part is in the correct position and detecting defects in each part, which supports the automation of assembly and final inspection stages in vehicle manufacturing.

03

Developed a model that achieves approximately 88% performance in the data processing sector for training line detection models essential for autonomous parking and driving, which minimizes labeling burdens.

Are you curious about
SNUAILAB's AI technology
applied across various industries?
Maximize your business values
by adopting industry-specific
AI solutions.
Are you curious about SNUAILAB's AI technology applied across various industries?
Maximize your business values by adopting industry-specific AI solutions.

   sales@snuailab.ai  sales
        snuai@snuailab.ai  general

Seoul  HQ
Seoul National University, 133 Building, Automation and Systems Research Institute, 208th Floor, 1 Gwanak-ro, Gwanak-gu, Seoul Special City

Gwanggyo  R&D / Business
1202th Floor, Building A, Gwanggyo Techno Valley, Advanced Institute of Convergence Technology, 145, Gwanggyo-ro, Yeongtong-gu, Suwon-si, Gyeonggi-do

© 2024 SNUAILAB Co., LTD. All Rights Reserved. 
Privacy policy     Terms and conditions

   sales@snuailab.ai  sales      snuai@snuailab.ai  general

Seoul  HQ  
Seoul National University, 133 Building, Automation and Systems Research Institute, 208th Floor, 1 Gwanak-ro, Gwanak-gu, Seoul Special City


Gwanggyo  R&D / Business  
1202th Floor, Building A, Gwanggyo Techno Valley, Advanced Institute of Convergence Technology, 145, Gwanggyo-ro, Yeongtong-gu, Suwon-si, Gyeonggi-do

© 2024 SNUAILAB Co., LTD. All Rights Reserved.   Privacy policy     Terms and conditions