Scrypted Docs

Appearance

Troubleshooting ​

Choose from the following troublehsooting guides. If your issue is not listed, please join Discord and report the bug in the #nvr channel.

Troubleshoot:

Object Detection ​

Detection Flow Chart ​

Object Detection failure is typically due to the camera hardware not properly supplying motion events. The camera's hardware motion sensor will trigger the object detection process when motion is detected, allowing the system to remain idle.

View the camera in the NVR and verify there are motion events in the Timeline. Detected Motion will be denoted with a thin and solid blue line, as seen below.

No Motion Detected
Detected Motion

If there is no motion visible on the timeline, the camera is misconfigured and not reporting motion properly. Enable the camera's hardware Motion Sensor via the manufacturer admin webpage or app. Ensure there are no zones filtering the motion.

In some cases, the camera motion sensor may be unreliable or non functional. Scrypted NVR provides software motion detection options to fill this gap. Enable a Motion Detection Extension on the camera.

If the motion sensor is functional, the Object Detection itself can be tested by dropping an image into it. Save the following image and dropping it into Scrypted NVR Object Detection:

Sample Image
Detected Objects

Recording ​

Recording issues typically stem from the storage directory or device not being properly configured.

  1. In the Scrypted NVR Plugin Settings, verify the storage directory is configured and is reporting the correct disk size and free space. The drive should ideally be at least 1 terabyte.
  2. Refer to the Recording Storage docs to properly format the drive with the correct filesystem and, if applicable, mount it into Docker.

Event Debug Tool ​

The event debug tool provides insight into what objects and motion was detected during an event. The tool is accessible via a desktop browser only. Click the camera label to go to the Event Debug Tool.

No Motion Detected
Detected Motion

Reducing False Positives ​

Object detection zones on the camera can be used to reduce false positives or remove uninteresting detections. Create a zone in the Object Detection Settings for the camera. The zone should accomodate the bounding box that may spill into the zone. E.g., the zone may need to be set back further than would seem intuitive since the bounding box for an object may be larger than may seem obvious:

Although the car is fully on the street, if a zone ends at the edge of the lawn, the car's bounding box would intersect the lawn.

An intersect zone will filter only detections that intersect (even a single pixel) the zone. A contain zone will filter detections that are completely encompassed by the zone.

The default zone type is include; a zone can be set to exclude to remove all detections of the matching type from being detected (such as problematic stationary cars).

High CPU Usage ​

Scrypted NVR will use considerable CPU and GPU while performing object detection on cameras that have reported motion. The following will help diagnose unusual system activity.

Verify GPU ​

Improperly configured systems may not be able to access the GPU when performing object detection. The official install scripts will typically set up the system correctly.

  • GPU usage for decoding can be verified by installing the Diagnostics Plugin and running System Validation.
  • GPU usage for detection can be verified by checking the OpenVINO or ONNX plugins.

Camera Activity ​

Cameras that are actively performing object detection can be viewed in the Video Analysis Plugin. There are two sections:

Active Motion Detection ​

There should generally not be any cameras that are using software motion detection. All cameras that are officially supported by Scrypted NVR have functional hardware motion sensors. If there are cameras listed in this section, please review the motion sensor setup to confirm that enabling software motion detection was necessary. The software motion sensor should be disabled.

Active Object Detection ​

There may be multiple cameras listed in this seciton if the high cpu usage is persistent. This may be due to cameras that are constantly reporting motion and never going idle. For example, trees blowing in the wind will trigger object detection. To resolve this issue, use motion zones on the camera to reduce unimportant motion that is reported to Scrypted. For example, zoning the camera to only perform motion detection on walkways and driveways, rather than on landscaping features.

Notification Delivery Failures ​

Complete Notification Failure ​

If notifications are not being delivered at all, this is likely due to an issue with the device itself. Perform the following steps:

  1. If the device is Android or iOS, reboot the device.
  2. Completely quit (swipe close, close tray, etc) and reopen the Scrypted NVR app to trigger a device reregistration.
  3. Search for the notification device in Scrypted Management Console and send it a test notification.
  4. Observe any errors in the Console.

No Images in Notifications ​

If notifications are working, but there are no images in notifications, this is likely due to a Scrypted Cloud plugin misconfiguration. Perform the following steps:

  1. Ensure the Scrypted Cloud plugin is installed.
  2. Log into the Scrypted Cloud plugin again.

This should resolve most notification image failures.

If the issue persists, and the server is behind a custom domain. The error is probably due to improper SSL termination or reverse proxy. To diagnose, delete and reinstall the Scrypted Cloud plugin to verify notification image delivery works on the Default settings.