i-PRO AI Applications
Install Various types of applications to keep your i-PRO network camera on the cutting edge.
There are many types of AI applications that can be installed on network cameras at no additional cost. Although i-PRO network cameras are edge devices, they are in charge of various detection processes. As a result, the camera is responsible for the advanced processing and the server does not require expensive hardware.
i- PRO AI applications
AI Video Motion Detection (AI-VMD)
AI differentiates between vehicles, motorcycles, bicycles, people, and objects detected by AI on-site learning. It further detects and sends warning notifications when they enter a specified area. It can also count the number of times the target object crosses line.
- It is possible to issue an alarm when a moving object enters a specified area.
- It is possible to issue an alarm when a moving object enters a specified area and stays there for a specified amount of time.
- It is possible to issue an alarm when an object moving in the specified direction crosses a specified threshold. (Counting is also available)
- Tailgating can be configured for alarm notification when multiple cross line are detected within a specified time period.
- It is possible to issue an alarm when a moving object enters a specified area and moves in the specified direction.
- It is possible to issue an alarm when a moving object exceeds a specified speed.
AI On-site Learning
AI On-site Learning allows to enhance AI analysis by adding target object and improving accuracy.
You can add customized objects to detect using on-site camera footage through simple learning process. In addition, it contributes to reduce missed/false alarms by training based on the past footage.
| Supported camera | WV-X15xxx, WV-X22xxx, WV-X25xxx (New X series Camera)
Note: To be used with AI-VMD |
| SD card | Not required |
| No. of learning images required | 10 – 200 images |
| No. of target to be added | Up to 5 targets
Note: Additional targets can be detected simultaneously |
* For learning process, collecting images and manual operations to teach the correct target are required. Also, detection performance depends on the learning results and the environments you use. Be sure to check the operation at the installation site.
* For configuration (detection area etc.), please use the latest version of i-PRO configuration tool (iCT).
AI Privacy Guard
To protect privacy and portrait rights, it is possible to automatically apply a mosaic the entire face and figure of a person captured by the camera.
| Privacy Guard target | Face, Human, Except face, Except human, All
* “Except face, Except human, All” are available with the firmware released in February 2024 or later |
| Stream | Stream(1), Stream(2), Stream(3), Stream(4), Stream(2)&(4) |
| Masking method | Blurring, Filling |
AI Sound Classification
AI detects and identifies (gunshot / yell / vehicle horn / glass break) sound. It then sends a warning notification according to the sound identified.
| Classification target | 1. Gunshot
2. Yell 3. Vehicle horn 4. Glass break |
| Classification condition | Target sound is 6 dB louder than crowd. Target sound continues more than a second.
(except for gunshot) |
AI Face Detection
The i-PRO network camera utilizing an AI engine and i-PRO Active Guard realize real-time face recognition without an expensive face recognition server. AI detection enables highly accurate face detection, and is also effective for detection in low lighting, backlit environments, or for detecting faces obscured by masks.
| Range of angle | Horizontal: 0° – 45°, Vertical: 0° – 30° |
| Detection number at the same time | Max. 8 faces |
| Size of face image | Min. size: width 30 pixel
Recommend: 60 pixel |
| Min. illumination level | 100 lux |
| Max. number of cameras | 20 when installed with VMS server PC
60 when installed in dedicated server PC |
| Max. number of watchlist | 1,000 faces |
| Recommended model | WV-S1136, WV-S1536LTN, WV-S2536LTN |
AI People Detection
Characteristics of the human face and clothing are extracted and classified into useful categories such as gender, age, and clothing color. This abundant attribute information can then be searched to easily find the target person.
| Attributes | Values |
| Gender | Male / Female |
| Age | Child (0-10) / Young Adult (11-20) / Adult (21-60) /
Senior (61+) |
| Mask | Mask / No Mask |
| Hair type | Long hair / Short hair |
| Hair color | Black / Brown / White / Gray / Gold |
| Hat | Hat / No Hat |
| Sunglasses | Sunglasses / No Sunglasses |
| Beard | Beard / No Beard |
| Top type | Long sleeves / Short sleeves |
| Top color | Black / Brown / White / Gray / Green / Red / Blue /
Yellow / Orange / Purple / Pink |
| Bottom type | Long / Short |
| Bottom color | Black / Brown / White / Gray / Green / Red / Blue /
Yellow / Orange / Purple / Pink |
| Bag | Bag / No Bag |
| Bag color | Black / Brown / White / Green / Red / Blue / Yellow /
Orange / Purple / Pink |
| Shoes color | Black / Brown / White / Green / Red / Blue / Yellow |
| Conditions | Range of angle | Vertical: 0° to 45° |
| Detection number
at the same time |
Max. 20 people | |
| Size of
people image |
Min. size: width 60 pixel
Recommended size : width 100 pixel |
|
| Min. illumination
level |
Min. illumination level : 10 lux
Recommended illumination level : 50 lux |
|
| Mask Area (non-detection area) | Max. 8 areas (polygonal) can be set | |
AI Vehicle Detection
Vehicle characteristics are extracted and classified into useful categories such as type and color. This abundant attribute information can then be searched to easily find the target vehicle.
| Conditions | Range of angle | Vertical: 0° to 90° |
| Detection number at the same time | Max. 20 vehicles | |
| Size of vehicle image | Min. size: 70 pixel for the shorter axis (about 15-20 m ahead)
Recommended size: 112 pixel for the shorter axis |
|
| Min. illumination level | 5 lux | |
| Mask Area (non-detection area) | Max. 8 areas (polygonal) can be set | |
AI Occupancy Detection
It counts the number of objects in a specified area and issues an alarm when the number exceeds a specified threshold. Information of the number of objects in the area can be output externally via HTTP or MQTT.
| Size of people image | Min. size: width 64 pixel (in case of S series 2MP model) |
| Detectable angle of view | Vertical: 0° to 45° |
| Detection object | People / Vehicles / Bicycles /
Objects detected by AI on-site learning |
| Max. number to detect at the
same time |
People / Vehicles / Bicycles: Max. 40 objects
Objects detected by AI on-site learning: Max. 20 objects |
| Detection Area | Max. 4 areas (polygonal) |
| Number of objects to detect | 1 to 40 objects/area |
| Detection time | 0.1 to 3,600 sec/area |
| Recommended location to install | People: Indoor
Vehicles / Bicycles / Objects detected by AI on-site learning: Indoor/ Outdoor |
| Min. illuminance | 50 lux |
It can define/learn the normal scene for each camera and output an alarm when it is determined to be abnormal.
- If the normal status is an empty area, the application will notify you when a stationary object is left behind.
- If the normal status is a closed door, the application will notify you when the door is left open for a defined period of time.
- If the normal status for a spot is set to “occupied”, the application will notify you when the spot is vacant.
- If the normal status is set to “full shelf”, the application will notify you when the shelf is empty or low, as defined.
| Scene Change judgment method | The user trains the app with multiple images of normal scenes, and the app monitors the amount of pixel change from
the trained images. It issues an alarm when the amount of pixel change exceeds a threshold for a specified time period. |
| Recommended location to install | Indoor |
| SD Memory Card | Required |
| Detection size | 5% or more of the width of the angle of view |
| Detection Area | Max. 16 areas (quadrilateral) |
| Detection time | 2 to 3600 sec/area |
AI Processing Relay
It can be used when you want to add an AI function to an already installed camera. This software receives MJPEG data from other cameras, performs AI processing with AI Extension Software, and performs alarm reception and image search with i-PRO Active Guard.
| WV-S71300-F3 (i-PRO Mini)
Note : Camera function on install camera is not available. |
WV-X15xxx, WV-X22xxx, WV-X25xxx (New X series Camera) Note : Camera function is available. | ||
| Input | Up to 1 ch | Up to 3ch | |
| No. of apps to be installed *1 | Up to 2 apps | Up to 3 apps | |
| Supported camera | i-PRO camera
Other Company Camera *2 |
||
| Received video format | Video format | H.265, H.264, JPEG *3 | |
| Resolution | Recommended resolution : 2MP(1920×1080) or 1MP(1280×720) Max. resolution : 1920×1080
Min. resolution : 640×360 |
||
| Available AI app *4 | AI-VMD / AI People Detection / AI Vehicle Detection / AI Occupancy Detection / AI Scene Change Detection | ||
| Supported System *5 | Video Insight / Genetec / Milestone / ASM300 / Luxriot with i-PRO Active Guard
/ Network Optix with i-PRO Advanced Object Search Plugin |
||
- For more details, refer AI apps compatibility in Accessory Selector https://i-pro.com/selector/accessory/ or the Technical Information [C0103] https://i-pro.com/products_and_solutions/en/surveillance/learning-and-support/knowledge-base/technical-information
- The camera must be able to output “Received video format”. And you need to know JPEG stream CGI or RTSP stream URL.
- [H.265]
- Only 1 non-AI camera can be connected.
- The frame rate of the connected non-AI cameras must be 3-10 fps. (Recommended 5 fps)
- The frame rate of non-AI cameras must be within 10 fps and the bit rate must be within 1 Mbps.
[H.264] - The frame rate of the connected non-AI cameras must be 3-15 fps. (Recommended 5 fps)
The total frame rate of the connected non-AI cameras should be kept within 15 fps, and the total bitrate should be kept within 3 Mbps.
Example: - When connecting 3 non-AI cameras with X-series camera: within 5 fps and 1 Mbps per channel
- When connecting 2 non-AI cameras with X-series camera: within 7.5 fps and 1.5 Mbps per channel
- When connecting 1 non-AI camera with X-series camera: within 15 fps and 3 Mbps per channel.
[JPEG] - The frame rate of the connected non-AI cameras must be 3-15 fps. (Recommended 5 fps)
- The AI analysis performance depends on the image quality, so it will be a best effort. Please check it with cameras you use on site.
- ASM300 with i-PRO Active Guard” supports only i-PRO cameras.
“Video Insight with i-PRO Active Guard”, AI Occupancy Detection alarms are not supported because they do not go through i-PRO Active Guard.
Supported functions with AI Processing Relay for each VMS Video Insight with i-PRO Active Guard
AI-VMD/AI People Counting for 360-degree fisheye camera
AI Video Motion Detection
- AI differentiates between vehicles and people, detecting and sending warning notifications when an intruder enters a specified area.
- Intruder detection:
Issues an alarm when a moving object enters a specified area. - Cross Line detection:
Issues an alarm when an object moving in the specified direction crosses a specified threshold. - Loitering detection:
Issues an alarm when a moving object enters a specified area and stays there for a specified amount of time.
People Counting
- Cross Line Counting:
Counts the number of people moving in a specified direction and cross a specified threshold. - Area Counting(Queue Management):
Counts the number of people in the set area.
Detection number at the same time is Max. 100 people.
Heat map
Heat mapping provides statistical information about traffic flow and keeps counts for people passing through and loitering in the area.
Occupancy detection
The network camera uses its AI engine to detect congestion, providing data that can be used to direct visitor traffic flow in advance or help staff work more efficiently.
AI Privacy Guard for 360-degree fisheye camera
AI Privacy Guard
AI can automatically mosaic faces and figures of people who are photographed to safeguard privacy and portrait rights.
i-PRO Active Guard
The i-PRO Active Guard stores the best images and metadata captured by i-PRO network cameras, then collates this data with the watch list registered in the client software and issues an alarm when a match is found. The server does not require expensive hardware because i-PRO network cameras handle the advanced processing. The server can even be installed on the same hardware as the VMS. The system comprises the i-PRO Active Guard, the AI application installed on i-PRO network cameras utilizing AI engines, and i-PRO Active Guard Plug-in software for the VMS client.
Example of a basic system connection
Specification
| Video Insight system | Genetec system | Milestone system | Luxriot | WV-ASM300W | ||
| Scale | Supported camera | i-PRO network camera with AI engine | ||||
| Number of camera | 1 to 100 CH when installed with VMS server (AI Face Detection is up to 20)
1 to 300CH when installed in dedicated server (AI Face Detection is up to 100) |
1 to 300CH when installed in dedicated server
(AI Face Detection is up to 100) |
1 to 4CH when installed with WV-
ASM300W (AI Face Detection is up to 4) 1 to 300CH when installed in dedicated server (AI Face Detection is up to 60) |
|||
| Number of client | No limitation (depending on limitation of VMS) | |||||
| Number of Recording server | 12 VI IP servers | – (only main server should be registered)
– * Only main server should be registered, Federated architecture is not supported |
– (Management server should be registered)
– * Management server should be registered, Federated architecture is not supported |
Luxriot Evo Server should be registered. | 12 Network disk recorders | |
| Number of
i-PRO Active Guard server per a client |
No limitation
(depending on capability of whole system) |
100 | ||||
| Supported AI
applications |
For plugin | AI-VMD / AI Sound Classification / AI Face Detection / AI People Detection / AI Vehicle Detection / AI Scene Change Detection / AI On-site Learning / AI Processing Relay / Vaxtor ALPR, ALPR MMC, OCR Genesis, OCR Container | ||||
| For dashboard | AI Face Detection, AI People Detection, AI Vehicle Detection, AI-VMD/AI People Counting for 360-degree fisheye camera,
AI-VMD, AI Occupancy detection, AI On-site Learning, Vaxtor ALPR, Vaxtor ALPR MMC |
AI Face Detection, AI People Detection, AI Vehicle Detection,
AI-VMD/AI People Counting for 360-degree fisheye camera, AI-VMD, AI Occupancy detection / AI On-site Learning / Vaxtor ALPR, Vaxtor ALPR MMC |
||||
| Store | Retention period | Max. 31 days limitation for face, people or vehicle data / Max. 92 days limitation for people counting data
* Max. 397 days (Image, Alarm) and 732 days (Counting data) by upgrading SQL Server Standard edition or higher |
||||
| Post Search | Filter | People, Vehicle (attribute, region of interest, date & time, camera, moving direction) Face (similar face, region of interest, date & time and camera)
LPR (License plate number, country, vehicle type, color, brand, model, moving direction, date & time, camera) OCR (OCR code, date & time and camera) Container (container code, size code, owner country / city / company), date & time and camera |
||||
| Similar search | People and vehicle (by same attribute information as detected ones),
Face (search for similar faces or region of interest), LPR (license plate number) and OCR (OCR code and container code) |
|||||
| Sort | Descending, Ascending, Relevance (Face, People, Vehicle, LPR and OCR) Timeline view (VI, Genetec and ASM300 only) | |||||
| Alarm | Watch list alarm | Up to 30,000 faces, Up to 12 people attributes, Up to 12 vehicle attributes Up to 12 License plate groups or attributes, Up to 12 OCR groups | ||||
| Detection alarm | AI Sound Classification (Gunshot / Yell / Vehicle horn / Glass break)
AI-VMD (Intruder / Loitering / Direction / Line cross / On-site learning)/ LPR, OCR (OCR code, container code) VCA, Access Control / AI Scene change detection |
AI Sound Classification (Gunshot / Yell / Vehicle horn / Glass break)
AI-VMD (Intruder / Loitering / Direction /Line cross / On-site learning) AI Scene change detection AI Occupancy detection LPR OCR (OCR code, container code) |
||||
| Related function | Show alarm on real-time, Alarm search,
VMS’s Rules configuration, Map overlay, Notification to mobile app, Manage alarm status |
Alarm destination (plugin, camera, IO Unit or Custom HTTP) and alarm trigger can be
configured. |
||||
| Playback/ Export | Playback | Playback video around the time of the best shot on full and multi-view | Playback video around the time of the best shot on full screen | |||
| Export video | Save best shot image, Export video from Recording server | |||||
| Export Search result | HTML and CSV | HTML | HTML | HTML | HTML and CSV | |
| Dashboard | Supported browser | Microsoft Edge, Google Chrome and Firefox | ||||
| Chart |
|
|||||
| Customize | Contents, display size, location for each chart.
3 Layouts per user can be saved and 24 user can be registered. (Up to 4 user can be log-in at the same time) Basic display color theme (dark or light) Line chart type (Straight line / Smoothed line), line/area name for counting |
|||||
System requirement
| VMS | WV-ASM300 | ||
| i-PRO Active Guard server installed in dedicated server
*Not support to install with Milestone XProtect |
Up to 100 cameras
*Camera with AI Face Detection is up to 60. |
|
|
| Up to 300 cameras
*Camera with AI Face Detection is up to 60. |
Microsoft® Windows Server 2016 Standard Microsoft® Windows Server 2019 Standard Microsoft® Windows Server 2022 Standard Microsoft® Windows Server 2016 Datacenter Microsoft® Windows Server 2019 Datacenter Microsoft® Windows Server 2022 Datacenter
|
||
| i-PRO Active Guard server installed with VMS server | Up to 100 cameras | • Intel® Xeon® Silver 4208 2.1 GHz (8 core 16 thread) or better
• 32 GB of RAM or more • 64 bit operating system Microsoft® Windows Server 2016 Standard Microsoft® Windows Server 2019 Standard Microsoft® Windows Server 2022 Standard Microsoft® Windows Server 2016 Datacenter Microsoft® Windows Server 2019 Datacenter Microsoft® Windows Server 2022 Datacenter • GbE network interface card |
|
| i-PRO Active Guard server installed with ASM300 | Up to 4 cameras |
|
|
Note : i-PRO Active Guard server can be run on Microsoft® Hyper-V with Windows Server 2022 and VMware® ESXi 7.0 Update 3. When using virtual platform, CPU, memory, network adapters, and storage may become bottlenecks.
Please consider in advance the allocation of sufficient CPU, memory, and network settings to meet the above hardware requirements.
Trademarks and registered trademarks
All other trademarks identified herein are the property of their respective owners.
Important
- Safety Precautions : Carefully read the Basic Information, Installation Guide and Operating Instructions before using this product.
- i-PRO Co., Ltd. cannot be held responsible for the performance of the network and/or other manufacturers’ products used on the network.
- When monitoring remotely, please pay attention to the network design.
- For details, please contact your local i-PRO sales representatives.
All images are for illustrative purposes only.
Specifications are subject to change without notice
https://i-pro.com/products_and_solutions/surveillance
Documents / Resources
 |
i-PRO AI Applications [pdf] User Guide WV-X15xxx, WV-X22xxx, WV-X25xxx, WV-S1136, WV-S1536LTN, WV-S2536LTN, AI Applications, AI, Applications |