1. VMS/NVR
| Overview
Provides all functions related to video such as IP camera control , video recording , video analysis , and video retransmission
| Background
When implementing the integrated SI (Building/Security Integrated Control) project, there were difficulties in interworking due to the passive
and closed structure of CCTV solution companies.
It is thought that video is an essential part of all security control in the future, so it was developed based on the judgment that it would be
better to develop it with our own technology rather than relying on an external VMS.
| Function
International standard protocol (ONVIF) support, convenient camera management functions : Device Discovery, Batch setting change
Simultaneous support for total recording and event recording
Provides status monitoring function for abnormal situations such as camera abnormalities and network errors
Minimize network load through proxy video server and video display server
By providing a proxy function, simultaneous video monitoring is possible without load even when multiple user cameras are connected
Can be linked with not only mobile devices such as smartphones, but also general IP phones
Possible to provide VMS solution optimized for large-scale CCTV video management and NVR service for small-scale video operation
| Sample Screen
[CCTV Registration]
[Creating CCTV operation screen]
[CCTV-Fire interlocking screen]
| Applications Field
- Responding to various security needs and smart security
- VMS required for large-scale control, NVR in small CCTV market
| Reference
- Jamsil Lotte World Mall (1,700EA)
- Lotte Center Hanoi CCTV (1,000EA)
- GS Home Shopping New / Old Building CCTV
- LGD Vietnam CCTV video control
- Ulsan Honorsville CCTV / LPR / Emergency bell
- Kyobo Life Insurance , Kyobo Data Center , Ulsan Hyundai Apartment, etc.
2. AI Video Analysis
| OverView
Using OpenVINO, the Intel Vision AI platform, AI image analysis algorithm and image processing technology are linked on the EZPrime platform to provide an AI Video Analysis Solution
| Background
Request for increasing requirements for AI-based deep learning video analysis and providing cost-effective solutions
Optimized for EZPrime platform configuration including image processing technology and interworking with AI-based Intel OpenVINO platform
| Function
Intel Vision AI platform interworking development within the EZPrime-VMS platform (AI platform and CCTV SW integrated)
OpenVINO provides algorithms optimized through Intel's pre-trained deep learning models.
Supports various deep learning engines: Caffe, TensorFlow, MXNet, ONNX, etc
Provides CPU-based deep learning inference function through support of various HW accelerators (simultaneous 16CH ~ 32CH analysis)
Object recognition processing speed is very fast by applying an optimized analysis system
Provides video analysis function linked with Intel OpenVINO AI image analysis platform and Intel Vision accelerator
Through video analysis event and CCTV SW linkage, the analysis result can be checked immediately in real-time and playback video
| Sample Screen
[Real-time analysis operation screen-intrusion, loitering, falling]
[Real-time analysis history management]
| Applications Field
- Smart image search(human detection) - Incident / accident , lost child etc
- Escalator accident prevention
- Automatic location search for specific objects ( carts , strollers , fixtures, etc. )
- Intrusion detection , abnormal behavior pattern analysis ( loitering , falling )
| Reference
- Unmanned PC room access image analysis
- Large shopping mall security manager smart video search , Detection of intrusion into unauthorized areas at specific times , loitering and falling down
3. Smart Video Search
| Overview
Provides a fast image search function using AI algorithm image analysis technology in conjunction with Intel's Vision AI technology, OpenVINO platform.
| Background
Existing intelligent video analysis technology is mainly limited to real-time events (intrusion, wandering, object detection, motion detection,
etc.), so the search function in actual video playback is not implemented properly.
Development of smart search function to identify people who move in the COVID-19 situation and quickly search for incidents and accidents
| Sample Screen
Various condition search setting function (target camera, search time, search period, similarity, etc.)
Through the search condition setting function, various search conditions are set in advance, and then selected and used
Analysis of video playback files stored in the storage server
Provides various methods for selecting image search targets (target selection by automatic human object detection, setting of human objects in images, etc.)
Provides image analysis results from various angles on a person through the attribute (clothing color, bag, etc.)
Even during search, you can click the extracted thumbnail image to check the search result in real-time or as a playback video
Provides a function to display the search and analysis target as a box on the video screen
Check the analysis result through real-time or video playback (VMS function interlocking) function for the search result. Video file download linkage
Provides detailed history list of analysis results. Possible to infer the movement of the analysis target by using the target camera and analysis time
| Smart video search technique
| Applications Field
- Search for accident targets in shopping malls, find the elderly with dementia, find lost children, etc.
- Work that requires workers to spend a lot of time searching for images
| Reference
- Lotte Shopping Mall - Smart Video Search
4. Intel OpenVINO
| What is OpenVINO Toolkit?
- A toolkit to accelerate development of high performance computer vision & deep learning inference into vision/AI applications used from edge to cloud. It enables deep learning on hardware accelerators and easy deployment across multiple types of Intel® platforms
- Inside Design of Intel OpenVINO Toolkit
| Benefits of OpenVION toolkit
- Challenges in Deep Learning
-
Why Intel OpenVINO toolkit?
: High-Performance, Deep Learning Inference
| AI Video Analysis using Intel Open VINO
OpenVINO is a toolkit with deep learning optimization and hardware acceleration functions. It provides algorithms optimized through Intel's pre-trained deep learning models, enabling high performance and fast solution development.
- AI-based video analysis integrated platform
- Supported Frameworks
-
Video analysis solution integrated into EZPrime platform using OpenVINO
- Using Intel OpenVINO, we completed the development of Video Analysis solution interlocking with the EZPrime platform, an integrated device management solution, and launched the EZ Vision AI image analysis system.