WServer AI User Guide | Waversa Systems

Connect and log in

Open the device address supplied by your installer. HTTPS requires a trusted certificate for that address. The website language selector is separate from the device interface, which currently offers Korean and English.

Select Log in and enter the ID and password issued to you.
On a shared computer, leave Remember ID and Keep me logged in unselected. Keeping a login does not provide an unlimited session.
If prompted to change the password, enter the current password, new password and confirmation, save, then log in with the new credentials.
Open Live Monitoring / Playback and check the camera list and video before changing camera or AI settings.
Log out when finished. If the login screen keeps reappearing, check whether the same account is being used on other devices and review its session status.

Connect and log in
Role	Available tasks
Administrator	Manage cameras, AI, system settings, PTZ, I/O, audio, USB backup, event acknowledgement and user accounts.
User	View live video, playback, searches, events and disk status, and change your own account. Missing configuration buttons may reflect your account permissions.

Check live video

The Live view under Live Monitoring / Playback is the starting point for checking camera connections and AI results. Changing the viewing layout does not change recording settings.

Select Live and choose a 1-, 4-, 9- or 16-tile layout.
Use the camera list or previous and next channel buttons to display the required video. Use one tile to inspect a channel in detail.
Open the tile’s ⋯ menu to inspect stream information. If AI results are missing, first check AI overlay visibility.
Open PTZ control only when you need to move a supported camera. Hold a direction or zoom button to move the camera; release it to stop.
To review an earlier scene, use Play recent recording in the ⋯ menu. It opens recording from approximately 30 seconds earlier. Select Live to resume current monitoring.

Example intrusion zone, object, path and IN/OUT overlays — Actual product video renderer with an example scene and detection data. People, vehicles, paths, counts, confidence and intrusion states are not real detections.
**Red zone and Intrusion label**Read the intrusion state of a named zone. This label is not a button.
**Object boxes and paths**Review object positions and tracked people’s paths alongside the image.
**IN / OUT**Cumulative barrier / people counts. The numbers in this example are synthetic.

Connect cameras and configure recording

Under Cameras, open Settings for the required channel and confirm the channel in the dialog title. Prepare the camera address, account and connection type supplied by the installer. An empty camera name displays the channel number.

Enter a camera name and enable the channel.
Choose the actual connection type: ONVIF, Waversa or RTSP. For ONVIF or Waversa, select a discovery result or enter the address. For RTSP, enter the stream URL provided by the camera.
For a new connection, enter the camera user and password. Leave both fields blank to retain the saved account.
For ONVIF, choose Main (original) or Sub (small view) from the stream options offered by the device and camera.
Choose Continuous, Schedule or Event recording and set the times as described below.
For PTZ, select ONVIF or RS-485 Pelco-D. For Pelco-D, match the I/O port role, communication speed and camera address to the connected equipment.
Select Save and wait for the device to finish applying the settings. After the saved message, check live video and reopen Settings to confirm the recording mode. If an error remains, correct the indicated fields and save again.

Connect cameras and configure recording
Recording mode	Configure and verify
Continuous	Select continuous recording while the channel is enabled. Confirm actual storage through disk status and recording playback.
Schedule	Click or drag one-hour cells for each day. Selecting a day name toggles that entire day. Finish with Save in the camera dialog.
Event	Enter 0–30 seconds before and 0–600 seconds after an event. Recording covers motion, AI detection and alarm events; it does not retain every moment as continuous recording does.

Actual WServer camera name, recording mode and video-source settings — Actual product UI showing example camera settings. camera.example.invalid is an example address; account fields are empty.
**Camera name and Use this channel**Set the display name and whether the channel is enabled. Disabling it stops its video, recording and analysis.
**Recording mode**Choose continuous, weekly schedule or event recording, with pre- and post-event time where applicable.
**Camera type and Save**Enter the source and required connection information, then save to register it.

Connect AI models and verify execution

Open AI settings for the channel under Cameras. A connection between cards is an actual analysis setting. Check both whether a card is connected and whether it is running.

WServer AI card connections and person appearance options — Actual product web UI with example data. Models, connections, running states and counts are illustrative; available features depend on the device.
**Person appearance card**Select a card to view its analysis options under Details.
**Highlighted destination card**Select the source card, then a highlighted destination. Connection changes are saved automatically.
**× on the left of a card**Disconnect the input to that card. On a device, the change is saved automatically.

Example: connect area intrusion detection

Use this sequence when the models are available on the device. Follow the parent and child connections offered by the screen rather than substituting an unrelated card for a missing model.

Confirm the camera channel in the dialog title.
Select Camera video, then select Person detection from the available connection candidates.
Select the connected Person detection card as the source and connect Person appearance. Continue with Person appearance → Object tracking management → Area intrusion detection.
Check the connection line and status at each step. A parent can feed multiple analyses, but only the offered connection candidates can be selected.
Select Area intrusion detection to open its detailed settings below. If prompted to define an area, continue with Add zone in the next section.
To disconnect a card, select × on its left. Clear selection only clears the selected source card; it does not delete a connection.

Read connection and runtime status

Read the included, subscription or purchase badge together with the runtime status. A connection line alone does not mean that entitlement and execution requirements are satisfied.

Read connection and runtime status
Status	What to check
Running	The camera, required parent models and execution conditions are ready. Check actual detection results in live video and events.
Off	Check whether the model is connected and whether any required detection targets are selected.
Not subscribed · Inactive	Use Subscribe or Manage on the card to review entitlement for the model and channel. Connecting it alone does not enable execution.
Upstream model not subscribed · Inactive	Check entitlement along the parent path first. Subscribing only to the downstream model may not resolve it.
Waiting / Waiting for NPU	Check whether the camera is enabled and connected, its parents, and the device analysis status. A saved connection is not proof that analysis is running.
Save error / Waiting for device response	Check connectivity and administrator access, then inspect the current card state again. Do not assume the last value you clicked was saved.

Detailed options and when they apply

Select a connected card to display its detailed settings below. Available options and execution depend on the installed models.

Detailed options and when they apply
Setting	Input and application
Object detection	Choose the object classes. Selection changes save automatically. If needed, draw and save an analysis region separately.
Person appearance	Choose available options such as upper clothing, lower clothing, hat, gender and age. If the screen reports that a model is unavailable, do not expect results for that analysis.
Object tracking management	Enter an integer from 1 to 1440 minutes for Object validity period and select Save. This retains object management information after the last detection; it is not a recording retention period.
Region, barrier and loitering dialogs	Edit inside the dialog and select Save. Their application differs from automatically saved card connections.

Configure analysis regions, intrusion zones and counting

An analysis region limits where a model analyses, an intrusion zone defines where intrusion is evaluated, and a barrier is a line crossed by people. Select the required model card before opening its editor.

AI settings with Add zone, confirmation time and the foot-position option — Actual product web UI with example data. The confirmation time and zone names are examples, not a recommended detection policy.
**Add zone**Open the editor for a new detection zone. Select an existing zone button to edit it.
**Confirmation time**Set how long the same tracked target must remain in the zone.
**Detect by visible foot position only**Choose whether to evaluate only targets with a detected foot position. Changes are saved automatically.

Actual zone editor with polygon vertices and a zone name — Actual product zone editor showing an example scene. The illustration, names and coordinates are illustrative, not a live camera image.
**Zone name**Enter a recognizable name, such as Restricted area.
**Image and numbered vertices**Click the image to add points and drag the points to adjust the boundary.
**Save**Apply the edited name and zone. Cancel leaves the current edit unapplied.

Define an analysis region

Use a model that offers Analysis region, such as Object detection or Person detection. The editor background is a recent recorded frame, not live video. If there is no frame, the editor reports a blank background; check the current camera view and recording status first.

Open Analysis region from the model card.
Click around the boundary to add 3–16 vertices. Avoid duplicate points or crossing edges.
Drag points to adjust them. Undo last point removes the final vertex. Whole picture removes the restriction and selects the entire image.
Select Save. If a vertex error appears, correct the shape and save again. Select Cancel to leave it unchanged.
Check the Selected region or Whole picture indicator in the detailed settings, then check live results in the intended area.

Configure area intrusion detection

First check the connected AI path and the runtime status of Area intrusion detection. An intrusion is reported when the same tracked object remains in the zone for the configured confirmation time.

Select Add zone on the Area intrusion detection card. Up to four zones are available.
Enter a Zone name, or leave it blank for the default. Draw 3–16 vertices on the frame and select Save.
In the detailed settings, enter an Confirmation time from 0 to 600 seconds. This change saves automatically.
Choose whether to enable Detect by visible foot position only. When enabled, objects without detected foot positions are not confirmed as intrusions, including scenes where feet are obscured.
Open an existing zone using its name button. To remove it, select Delete zone and then Save. Cancel leaves that deletion unapplied.
Check the zone name and active intrusion indication in live video, then inspect the camera and zone details in Events. Connect an output in Event Map if an alarm or speaker must operate.

Set barrier direction and check IN/OUT

Connect Barrier / people counting under Object tracking management and open Edit barrier.
Drag endpoints A and B into place. Crossing towards the arrow counts as IN; crossing the opposite way counts as OUT. Use Swap direction to swap them.
Choose whether to show the barrier and select Save. If the line is too short, move the endpoints farther apart.
Compare crossings with the live IN/OUT counters. If direction is reversed, reopen Edit barrier, reverse it and save.
Use Reset counts only when the accumulated totals should start again, and wait for the pending message to clear.

Adjust loitering confirmation time

Open Settings on the Loitering detection card.
Enter a whole-number Detection time (seconds) from 10 to 3600.
Enable cross-camera appearance linking only when evaluating that experimental feature at the site. Keep it distinct from ordinary single-camera configuration.
Select Save, then check card status and events. Cancel does not apply changes made in this dialog.

Review events and open recordings

The Events tab under Events lists recent activity. To investigate a specified earlier time, use Search as described in the next section.

Narrow the list with the event message search, camera and All states, Unread or Acknowledged filters. Filter changes update the displayed list.
Open an event picture to inspect its time, source, type, message and acknowledgement state. If No picture appears, review the text record first.
Select Play recording for a camera event to open single-channel playback approximately five seconds before the event. Playback requires a retained recording for that period.
After reviewing it, an administrator can select Acknowledge and check the completed state. Merely opening or closing the event does not acknowledge it.
To rule out hidden results, clear the search text and return to All cameras and All states.

Review events and open recordings
Situation	What to check
No picture or recording will not play	Check the event camera and time, recording mode at that time, disk status and retained recording range.
Disk event storage delayed	Check storage and query again. An empty list while this warning is present does not establish that no incident occurred.
Acknowledgement fails	Check administrator access and connectivity, then reload to see the current acknowledgement state.

Search conditions and recording playback

Search finds saved events, AI detections or recording thumbnails and opens playback for the relevant camera. Begin with broad conditions, then narrow the camera, time and analysis filters based on the results.

Search mode, date and camera filters with example event results — Actual product UI with example search results. Scene illustrations, channel names, times and events are illustrative.
**Events, Motion, AI detection and Thumbnails**Choose the search method for the records you need.
**Date and time, camera and Search**Set the time range and channel, then run the search.
**Result card**Select a scene to open recording playback from five seconds before the event.

Choose a date and time range

Open Search and select the date and time button.
Choose a calendar day and enter start and end times. Search a period within one day; change the date to investigate additional days.
Select Apply. The selected period appears on the button and the query starts. Times use the device’s configured time zone.
Choose a camera and search mode. After changing camera or other filters, select Search to refresh. Changing the search mode or AI detection kind automatically runs a new query.
Read the result count and the message below the results. If asked to narrow the range or wait while records are read, do not treat the current list as the final investigation result.

Choose a search mode

Choose a search mode
Mode	Purpose and use
Events	Find camera and device events in the period. Camera results open playback about five seconds before the event. A device-wide alert may have no camera playback target.
Motion	Filter recorded motion events. Motion appearing in video alone does not create a result; detection and recording had to be configured at that time.
AI detections	Select an available detection kind. Person-related kinds can offer upper/lower observed colour, hat, other appearance filters and track ID.
Thumbnails	Select one camera and a 1-, 5-, 10-, 30- or 60-minute interval to scan scenes. All cameras uses the first camera in this mode, so choose the intended camera explicitly.

Play a result and adjust position

Select a camera result picture or card. Playback opens that camera in a single tile.
Use Play, Pause and Reverse. Select x2, x4, x8 or x16 for speed; selecting the active speed again returns to x1.
To jump to an exact time, open the calendar, choose a date and hour, minute and second, then select Go.
Drag the playback position bar to inspect a time and release it to seek. Changing cameras or layout during playback selects recordings for the displayed channels.
Use USB backup for a required segment. Select Live when finished to return to current monitoring.

When results or pictures are missing

When results or pictures are missing
Symptom	Next check
No results	Confirm camera, date and device time zone; remove additional appearance filters and search again. Check whether AI, motion detection and recording were active at that time.
Too many results	Narrow the period or detection conditions. Do not assume that the displayed list contains every result.
Some records could not be read / index still loading	Query again later and check storage. Distinguish a read failure from no matching events.
Blank thumbnail cells	Wait for picture loading, then check recordings at that time. If cells remain blank, check recording mode, retained range and connection errors.

Connect events to displays, alarms and speakers

Event Map defines which outputs operate when a source event occurs; it does not configure the detector itself. First confirm that the camera and AI are running and events are being recorded.

Select a detection source grouped under its camera in Event Map.
Select a screen, alarm, speaker or other available output. Check the connection line and count. The change saves automatically.
While the same source is selected, clicking an already connected output removes that link. An unlink button can also remove a selected connection or all connections from the source.
To connect one source to multiple outputs, select the outputs in turn. If several sources share an output, any one of those sources can trigger it.
For additional integration, use Add output to configure an available server notification or installed script output with the destination and settings approved by the installer.
Use an output test only after confirming the operating scope with the site operator. If saving fails, recheck which connections actually remain.

Configure I/O and voice announcements

Use I/O Settings to configure the serial ports, alarm contacts and speaker functions actually available on the device. Use electrical and communication settings confirmed by the installer.

Match the serial-port role and communication settings and alarm contact types to the connected equipment, then select Save in I/O Settings.
Under Speaker, select Add clip and choose a built-in sound or a WAV file up to 4 MB. Set its name and add it.
Confirm the clip appears, then adjust its volume and repeat count. Connect it as an output in Event Map to play it on detection.
For background music, choose a prepared folder and track, then use Play, Pause, Stop and Volume. Event alarms pause the background music before it resumes.
Check the microphone destination and allow browser microphone access. Hold PTT or Talk while speaking; release it to end transmission.

Configure networking and external access

Use Network, DDNS and P2P registration / management under System Settings. Before changing the address, prepare the new address and tell other operators that their connection may be interrupted.

Open Change for Network and choose Automatic (DHCP) or Static IP.
For a static address, enter the address, subnet mask, gateway and DNS values supplied by the installer. Save, then reconnect at the new address and check status.
Open P2P registration / management and check both connection status and service status. Being connected and having an approved, current entitlement are separate conditions.
If registration is authorised, enter the Purchasing organization number and PIN, then select Register and issue URL.
Check the completion message and connection URL. Use Copy URL, or copy the selected value manually if clipboard access is unavailable.
Open the issued URL from the external network, log in with the device account and verify the required video. Check DDNS separately where the installation uses it.

Manage accounts, time and storage

System Settings contains your account and device operating information. Regular users see only permitted items such as their own account.

Open Change for Account, enter the current password and required new ID, password and confirmation, then save. Log in again with the new credentials after the account changes.
Administrators can enter a user ID and a password of at least eight characters under User accounts and select Add user. Confirm the target user before resetting a password or deleting an account.
Check that the time zone matches the installation and inspect time synchronisation under NTP settings. Start here if event or recording timestamps appear incorrect.
Inspect disk connection, capacity, recording status and warnings. The web disk screen is for status and does not offer formatting.
Choose Fill (Crop) or Whole picture (LetterBox). Use Whole picture when the edges matter, and change the interface language if required.

Back up recordings and update firmware

USB backup writes recordings to a drive attached to the device. It does not download them to the computer running the browser.

Export a recording segment to USB

Attach a backup USB drive to the device and open USB backup from Monitoring or Search.
Choose the start time and enter a length in minutes or hours. Check the displayed end time and selected cameras. Use Select all or Clear all to change channel selection.
Select the destination in the USB list. Use Refresh USB if a new drive is missing. Recording disks are excluded as backup targets.
Choose MP4 or Waversa encrypted + Windows player. For a protected Waversa backup, enter both an ID and a password of at least eight characters; leave both blank for an unprotected backup.
Use Estimate size to check space, then select Start backup. Monitor progress and output size until Backup complete appears.
Select Eject USB and wait for permission to remove it before unplugging. Check that the required files are present on another computer.

Check and install an update

Open Firmware update under System Settings, check the current version and select Check for updates.
If a release is available, read its version and notes. Schedule installation for a period when recording can be interrupted.
Select Install and review the interruption notice. Monitor Downloading, Verifying signature and Installing.
If the web connection drops, wait for the device to reconnect. Log in again and check the current version, camera live video and recording status.
If Failed or Rolled back to previous version appears, record the version and status detail and consult the operator responsible for the device. Do not repeatedly install without checking the outcome.