Konan Watcher

AI-based Video Understanding Solution

Feature 1

UX innovation through vision AI technology

'Contents Management' innovation through video object detection

  • Face/object detection, situation awareness, voice recognition
  • Timeline-based metadata viewing
  • Intuitive post-calibration of metadata

AI-supported quick subtitling/quick dubbing

Multimodal search

Automatic creation of highlight videos

Feature 2

High quality solution built on 20 years' experience

Video understanding framework based solution built on 20 years' experience

Security Certification & GS Certification acquired

  • Intelligent CCTV evaluation certification (2021-12-09. defense business field)
  • GS Certification (expected to be acquired)

International standards applied

  • Metadata management based on the international Dublin Core standard
  • Operating without additional plug-ins by applying HTML5 standard

Feature 3

MLOps-based operation

MLOps-based automation of the entire deep learning training process

  • Automatic process of data collection-cleansing-training-validation
  • Training supported for voice/object/person detection in deep learning

Support for stable system operation

  • Active-active duplex configuration
  • Non-disruptive server patching with core and business logic modules separated

Intuitive web-based tools supporting operation tasks

  • Metadata field editor for customized services
  • Dashboard for integrated real-time monitoring management
  • Drag-and-drop workflow manager

Feature 4

Real-time video recognition

Processing and management of massive multimedia data processing (more than 100 million files)

  • On the fly transcoding implemented once video is acquired
  • High-speed transcoding utilizing GPUs
  • 30x speed cataloging

Real-time video recognition without delay

  • Object detection, face/voice recognition, anomaly detection

High-speed 8K video transmission using the data file partitioning technique

Feature 5

Flexible backbone system integration

High level/low level restful APIs enabling backbone system integration for each industry

  • Support for integration to drones and CCTV systems
  • Support for integration to LMS and video conferencing/webinar systems
  • Support for integration to video servers, news information systems, transmission systems and NLE systems

Support for integration to organizational chart for authorization management

  • Linking to Active Directory/Open Directory

Key Functions

System Architecture

Benefits

Video Value Enhancement using AI

  • Video Value Enhancement by tagging persons and objects and creating metadata

Improvement of Video Production Process using AI

  • Subtitle creation, dubbing, automatic creation of highlight videos, etc.

Increase in Video Reuse Rate by utilizing neural search

  • Improving responsiveness (e.g. subscriptions, views, reactions and comments) by increasing the density of video contents (including tags, subtitles, and dubbing)
Case study 1

'Konan Watcher' National defense

  • Identifies enemies and allies by analyzing data from videos and photos using deep learning technology
  • Building battlefield scenarios suitable for military command and control systems

Case study : 
'Konan Watcher' National defense

Konan Technology

Industry : National defense / Period : 10 months / Project Overview : AI-based military intelligent information platform

Case study 2

'Konan Watcher' SK Broadband

  • Provides personalized search and recommendation services by using deep learning technology to analyze deep metadata in contents such as movies, dramas, and shows
  • Automatically provides key highlight videos such as eating and action scenes

Case study : 
'Konan Watcher' SK Broadband

Konan Technology

Client : SK Broadband / Period : 8 months / Project Overview : AI-based DeepMeta Solution

Case study 3

'Konan Listener' Korea National Open University

  • Subtitles online courses for students including those with hearing loss
  • Creates accurate subtitles using speaker adaptive acoustic modeling to improve voice recognition rate

Case study : 
'Konan Listener' Korea National Open University

Konan Technology

Client : Korea National Open University / Period : 12 Months / Project Overview : AI-based subtitling system

Case study 4

'Konan Vision AI' Missing Person Search

  • A high-performance real-time object (person and face) recognition engine specialized for edge devices
  • Applicable for missing person searches in disaster and safety fields

Case study : 
'Konan Vision AI' Missing Person Search

Konan Technology

Case study 5

'AI-based Fire Operation System' ROK Army Training & Doctrine Command

  • Provides swift and accurate decision-making support in various battlefield scenarios
  • Enables optimal firepower operation tailored to target characteristics through a deep learning-based firepower recommendation engine

Case study : 
'AI-based Fire Operation System'

ROK Army Training & Doctrine Command

Konan Technology