AI Development

In the time of transformation into AI and machine learning, the coverage of computer vision technologies
has been expanded. To detect humans, things, behaviors, etc. and interpret their meaning, computer vision
technologies have fast grown with deep learning techs developing.

INFINIQ helps customers transform into AI through research and development (R&D) of computer vision
technology based on experiences and insights that it has gained while processing datasets of autonomous
driving, smart mobility, robotics, smart retail, smart appliances, factory automation, security monitoring, etc.
in many industrial fields.

Services Provided

INFINIQ’s AI lab, AI Studio, is dedicated to the R&D of computer vision technology.
In the entire territory of dataset workflow required for AI training,
it develops solutions and the vision recognition technologies necessary
to detect various objects and recognize behaviors.

R&D Areas

The lab researches and develops technologies from DataOps for ML/DL datasets
to processing and detecting 2D/3D sensor data for recognition by computer vision.

  • A system of DataOps to connect
    dataset processing for AI training
    with algorithm developing     Dataset workflows and insights optimized
    to utilize AI are provided to continue to supply
    datasets and upgrade models through collaboration.
  • Effective dataset design,
    characteristics sorting, and
    meta-analysis necessary for AI training     To enhance the inference performance of ML/DL,
    it is necessary to design dataset specs and auto-extract
    and analyze metadata.
  • Computer vision-based object detection,
    face recognition, behavior recognition,
    and tracking technology     In images and videos with deep learning, objects
    are detected, faces are recognized, and abnormal behaviors are detected, and they are tracked (HEX Vision).
  • 2D+3D sensor fusion
    and multi-modal calibration     Calibration, synchronization, and automatic object sorting
    to gain multi-sensor datasets for autonomous driving
    and smart mobility
  • 2D image
    / 3D LiDAR (point cloud) labeling     AI-based labeling tool and labeling automation
    to develop GT for algorithm learning
    by computer vision recognition
  • Data augmentation and synthetic data
    processing to create, transform,
    and synthesize data     Data augmentation and synthetic data processing
    under specific conditions such as edge case, corner case,
    abnormal case, etc.
  • Automatic deidentification
    of personal information in a video     A target that needs the protection of his or her personal data included in a video is automatically deidentified.

Results from R&D

AI Studio, the lab of INFINIQ, has succeeded in developing technologies available
for application in the fields of computer vision and data processing.

  • AI Engine
    in Computer Vision     Computer vision–based AI engine
    to recognize behaviors, objects,
    and faces
    HEX Vision
  • 2D/3D Labeling Tool An annotation tool with pretreatment
    applied by automatic object recognition and boundary extraction
    MYCROWD
  • AI Training Dataset
    Management Solution     An integrated dataset platform
    available to control annotation tools
    and resources, monitor performance,
    and manage datasets
    MYCROWD
  • Smart Shop Solutions A vision-based retail solution that may be applied from small
    unmanned shops to middle and large
    unmanned counter
    AI Counter