Release Notes

This section covers what changed in each version of VUNex AI.

1.0.0 (2026-09-03)

Release notes are provided from this release onward.
VUNex AI analyzes camera video in real time to detect intrusion, falls, fire and more, and runs the actions you assign when an event occurs.

You can teach the system the targets you want or build events by combining your own conditions, and detected objects can be searched later.
Access it from a web browser.

You can sign in from a web browser. On first access you change the default password and set security questions, and you can reset the password with those questions later.
You can select Korean, English, Japanese or Spanish on the sign-in screen.
You can set the screen theme to dark or light. The product is optimized for Google Chrome, and a guidance screen appears on unsupported browsers.

Home

You can review the day at a glance with today's events, the object detection trend and the event feed.
You can check the number of online, offline and error devices in device status, and how much analysis capacity is used in the AI Power widget.
You can view event status as a map, a list or statistics, and see the statistics by ROI and by hour.

Live View

You can monitor multiple cameras in a grid, and the auto grid applies the layout that fits the number of devices.
You can drag cameras into the video area and group them into collections that can be saved, edited and deleted. Up to 20 collections can be created.
You can show the camera list as a list or as thumbnails, and toggle object boxes, skeletons, detection zones and the camera name on the video.
The cell blinks when an event occurs, and you can open the latest event list and its details from the cell. The event tab on the right shows the notification list with thumbnails and the actions that ran.

Playback

You can query and play recorded video by camera and by period, drag the timeline to move to a point in time, or search by specifying a date and time.
You can select several videos and play them together in multi view.
You can use playback speed control, 5-second skip, reverse playback, snapshots and event markers, and start a similar person search directly from the playback screen.
You can choose a recording mode that stores only the segments where events occurred. In this mode the recorded video screen is organized around the event list.

You can search detected objects by period, device and accuracy.
You can turn text search, attribute search and similar person search on or off per object type.
You can combine object, event, gender, top, bottom, hat, carried item and color as detailed conditions, and upload an image containing your target to find scenes with similar objects.
You can select a thumbnail in the results to see the video clip and the location on the map.

AI Teaching

You can teach the system by selecting your target directly in the video. The teaching screen opens from both live video and playback, and you can move or resize the selection box.
You can rename a class you created later.

AI Event

You can use intrusion, fall down, loitering, theft, similar person match, class recognition, looking around, gaze detection, fire, forklift overspeed, wrong-way movement, concealing and line crossing.
You can build custom events by combining the conditions you want. Absence and equipment scenarios, condition combinations and overlap thresholds can be configured, and the events you create attach to detection zones and actions just like any other event.
You can exclude people riding a car, motorcycle or forklift from intrusion and loitering notifications, and turn this on or off per detection zone.
You can add, edit and delete actions per event and assign cameras and detection zones to each action. Action types include cloud transfer (image/video), cell alarm, pop-up message, thumbnail and ONVIF metadata.
You can schedule the actions that run when an event occurs by day of week and time slot. Frequently used time slots can be saved as presets, and the name and times of a saved preset can be edited later.
You can create several named detection zones per camera and turn them on or off per action. If none is set, the whole frame is the detection zone.
You can place a reference zone on the floor and enter the real distance so speed can be calculated for overspeed detection.
You can register HTTP custom actions to forward events to external systems. User-defined actions can also be sent to the cloud, and the delivery result is kept in history.

AI Service

You can configure the search mode per object type in the search service. A mode that is off stores no data.
You can upload single-floor or multi-floor maps in the map view service and place camera positions and fields of view.
You can preselect the classes used by events and services as interested classes. The AI model is changed in the App Store.
You can check the applied AI model, the classes it provides and AI Power usage in the App Store, and turn installed AI Events and AI Services on or off.
The models and modules an AI app needs are installed automatically, and you can also download an AI app as a file and install it directly. Installation and applied status is shown on screen in real time.

Device

You can register devices in four ways: auto scan, manual scan, manual add and file upload.
You can choose the RTSP transport among automatic, TCP, UDP and multicast, and assign live, recording and AI analysis profiles separately.
You can export the device list to a password-protected file or delete all devices at once, and enable cloud and message server integration.
You can configure the storage usage limit, retention period, thumbnail quality and log retention period, and check the estimated retention with the preview.

System

You can update the software by selecting a version from the cloud or uploading an update file.
You can perform factory reset, system reboot and shutdown with the administrator password, and view and copy the MAC address and the HTTP API bearer token.
You can select the time zone and set the time automatically via NTP or manually.
You can query the event log and the system log by period, device and type, and export them to CSV.
You can configure the network with DHCP or a static IP, check IP availability and set DNS.
You can choose HTTP or HTTPS for device access. Register a certificate to reach the screen over an encrypted connection.
You can export device settings to a file and load them onto another device, so you do not have to enter the settings again when installing several identically configured devices.
You can run VUNex AI on a standard PC with an NVIDIA graphics card, and check and install the required drivers and components in one step.

Offices and Technical Support

Office Locations

Headquarters: 서울특별시 금천구 디지털로 9길 47, 9층·11층 (가산동, 한신IT타워 2차)
Anyang Center: 경기도 안양시 동안구 시민대로 109번길 62

Technical Support

Email: PR_sales_team@innodep.com
Fax: 02-6336-3368
Call center: 1661-6806 (24 hours, all year round)

Reference