On Guard Security Assistant V4.0.0

This application works with almost any IP/network camera and adds artificial intelligence designed to identify security related pictures. More specifically, it uses DeepStack as an AI to intelligently identify objects in still (.jpg) pictures. It allows the user to monitor as many cameras as desired. It allows the user to visually create multiple areas that are of security interest, including the types of objects that are of interest in each area.

Version 3.0.1 and beyond (4.0) is a significant update to previous versions. The most dramatic difference is that it fully supports any known camera that supports ONVIF or HTTP. On Guard can now obtain and analyze pictures from (1) the camera itself without disk storage (2) messages sent by a camera using FTP, and taking advantage of motion sensing capabilities of the camera (3) Blue Iris or other applications that write motion related still images to disk.
On Guard 4.0.0 features
It is provided free of charge.

ONVIF and HTTP based cameras are fully supported. This should include almost any camera not locked into a manufacturer proprietary system.

It is fully compatible with, but not reliant on Blue Iris.

There is support for multiple cameras.

Support for cameras with different resolution and cameras from different manufacturers.

You can design areas of interest (zones) to identify or ignore activity in an area, including overlapping areas.

You can design your areas to be in almost any shape desired (a 64x64 grid on the pictures).

You can choose to base security notifications based on the size of identified objects (how big and how close they are to the camera).

Supports facial recognition.

Optionally manage the start/stop/reset of the DeepStack AI.

Notify any application that can take an action as a result of an HTTP message/request.

Supports IFTTT activation of lights and other smart devices.

Supports MQTT.

Supports email notification of clients with attached pictures. This also includes the ability to send MMS/SMS messages to phones that support it.

Supports Pan/Tilt/Zoom, and move to preset for cameras that support these features.

Tracks pictures with identified items of interest in a database so that you can move only through those pictures if desired.
Provides a timeline scroll bar. You can now use a track bar to scroll to pictures at any desired date/time.

Cleanup old/uninteresting pictures for one or all cameras.

Detailed documentation/instructions are included.

Can display snapshot pictures/video as an aid in defining areas of interest and camera positioning.
Features Not Supported
Does NOT store video to disk. However, it can trigger another application to start recording.

Does NOT playback videos.

While you can play live video (motion .jpg) directly from a camera, this capability is limited (and not smooth).

Does not support many doorbell cameras, and cameras locked into a proprietary camera ecosystem.

Version 4.0 Change List
On Guard no longer uses SQL Server. Instead, it uses Sqlite. This simplifies setup, and it eliminates the potential for any interference with existing SQL Server installations. Because this is a major change it warrants a change to the next On Guard version (4.0.0). On Guard 4 will no longer see motion identified in previous versions.

The menu system has been reorganized.

You can now set the log level to multiple values. This should help reduce the log spew as well as allow the user to focus on errors and problems.

You can now go to a specific picture date/time. In addition, you can use that date time as a fixed point if you want to see the same time period on multiple cameras.

Test pictures can be sent to individual cameras, rather than all cameras.

The cleanup of old pictures has been modified to speed it up.

Bug fixes and a lot of internal code changes.

On Guard Security Assistant V3.1.1

3.1.0
52b9b2a

This application works with almost any IP/network camera and adds artificial intelligence designed to identify security related pictures. More specifically, it uses DeepStack as an AI to intelligently identify objects in still (.jpg) pictures. It allows the user to monitor as many cameras as desired. It allows the user to visually create multiple areas that are of security interest, including the types of objects that are of interest in each area.

Version 3.0.1+ is a significant update to previous versions. The most dramatic difference is that it fully supports any known camera that supports ONVIF or HTTP. On Guard can now obtain and analyze pictures from (1) the camera itself without disk storage (2) messages sent by a camera using FTP, and taking advantage of motion sensing capabilities of the camera (3) Blue Iris or other applications that write motion related still images to disk.

On Guard 3.0.1 features

It is provided free of charge.
ONVIF and HTTP based cameras are fully supported. This should include almost any camera not locked into a manufacturer proprietary system.
Is fully compatible with, but not reliant on Blue Iris.
Supports multiple cameras.
Support for cameras with different resolution and cameras from different manufacturers.
You can design areas of interest (zones) to identify or ignore activity in an area, including overlapping areas.
You can design your areas to be in almost any shape desired (a 64x64 grid on the pictures).
You can choose to base security notifications based on the size of identified objects (how big and how close they are to the camera).
Supports facial recognition.
Optionally manage the start/stop/reset of the DeepStack AI.
Notify any application that can take an action as a result of an HTTP message/request.
Supports IFTTT activation of lights and other smart devices.
Supports MQTT.
Supports email notification of clients with attached pictures. This also includes the ability to send MMS/SMS messages to phones that support it.
Supports Pan/Tilt/Zoom, and move to preset for cameras that support these features.
Tracks pictures with identified items of interest in a database so that you can move only through those pictures if desired.
Provides a timeline scroll bar. You can now use a track bar to scroll to pictures at any desired date/time.
Cleanup old/uninteresting pictures for one or all cameras.
Detailed documentation/instructions are included.
Can display snapshot pictures/video as an aid in defining areas of interest and camera positioning.

Features Not Supported

Does NOT store video to disk. However, it can trigger another application to start recording.
Does NOT playback videos.
While you can play live video (motion .jpg) directly from a camera, this capability is limited (and not smooth).
Does not support many doorbell cameras, and cameras locked into a proprietary camera ecosystem.

Version 3.1.0 Change List

Object width/height is now stored as a percentage of the picture width/height rather than in terms of pixel sizes. This allows the picture resolution to change without affecting object matching. This is a breaking change for any areas that use the width/height feature. You will need to delete the area objects and re-enter them.
The format of email notifications has changed slightly to make it more compatible with many phone MMS displays. MMS phone applications are not all the same, and may change over time!
The pace of emails sent to email servers has been slowed down due to issues of some email servers balking at sending too many emails too quickly.
The Threads option for DeepStack has been made optional (0 = parameter not used). This is necessary because DeepStack version prior to 1/1/2022 don’t startup with this parameter.
Bug fixes.
NOTE: The DeepStack January 2022 release does not find any objects for some computers/graphics cards using GPU mode. This problem is not an On Guard issue. This issue has been reported to the DeepStack developers. Hopefully, it will be fixed soon. You may need to use the September 2021 release if you want to use the GPU mode. Also, note that performance should be improved substantially using 2022 DeepStack versions using the Threads option (see Application Settings). So, if the January 2022 DeepStack version works for you, set the Threads option and use it.

Version 3.1.1 Change List
Update the Release setup code to match the source code (it was out of date)
Added some additional error checking.
Revised the email format to make it more compatible with some email clients.
Some UI improvements.

On Guard Security Assistant V3.1.0

This application works with almost any IP/network camera and adds artificial intelligence designed to identify security related pictures. More specifically, it uses DeepStack as an AI to intelligently identify objects in still (.jpg) pictures. It allows the user to monitor as many cameras as desired. It allows the user to visually create multiple areas that are of security interest, including the types of objects that are of interest in each area.
Version 3.0.1 is a significant update to previous versions. The most dramatic difference is that it fully supports any known camera that supports ONVIF or HTTP. On Guard can now obtain and analyze pictures from (1) the camera itself without disk storage (2) messages sent by a camera using FTP, and taking advantage of motion sensing capabilities of the camera (3) Blue Iris or other applications that write motion related still images to disk.

On Guard 3.1 features

1. It is provided free of charge.
2. ONVIF and HTTP based cameras are fully supported. This should include almost any camera not locked into a manufacturer proprietary system.
3. Is fully compatible with, but not reliant on Blue Iris.
4. Supports multiple cameras.
5. Support for cameras with different resolution and cameras from different manufacturers.
6. You can design areas of interest (zones) to identify or ignore activity in an area, including overlapping areas.
7. You can design your areas to be in almost any shape desired (a 64x64 grid on the pictures).
8. You can choose to base security notifications based on the size of identified objects (how big and how close they are to the camera).
9. Supports facial recognition.
10. Optionally manage the start/stop/reset of the DeepStack AI.
11. Notify any application that can take an action as a result of an HTTP message/request.
12. Supports IFTTT activation of lights and other smart devices.
13. Supports MQTT.
14. Supports email notification of clients with attached pictures. This also includes the ability to send MMS/SMS messages to phones that support it.
15. Supports Pan/Tilt/Zoom, and move to preset for cameras that support these features.
16. Tracks pictures with identified items of interest in a database so that you can move only through those pictures if desired.
17. Provides a timeline scroll bar. You can now use a track bar to scroll to pictures at any desired date/time.
18. Cleanup old/uninteresting pictures for one or all cameras.
19. Detailed documentation/instructions are included.
20. Can display snapshot pictures/video as an aid in defining areas of interest and camera positioning.

Features Not Supported

1. Does NOT store video to disk. However, it can trigger another application to start recording.
2. Does NOT playback videos.
3. While you can play live video (motion .jpg) directly from a camera, this capability is limited (and not smooth).
4. Does not support many doorbell cameras, and cameras locked into a proprietary camera ecosystem.

Version 3.1.0 Change List

1. Object width/height is now stored as a percentage of the picture width/height rather than in terms of pixel sizes. This allows the picture resolution to change without affecting object matching. This is a breaking change for any areas that use the width/height feature. You will need to delete the area objects and re-enter them.
2. The format of email notifications has changed slightly to make it more compatible with many phone MMS displays. MMS phone applications are not all the same, and may change over time!
3. The pace of emails sent to email servers has been slowed down due to issues of some email servers balking at sending too many emails too quickly.
4. The Threads option for DeepStack has been made optional (0 = parameter not used). This is necessary because DeepStack version prior to 1/1/2022 don’t startup with this parameter.
5. Bug fixes.
6. NOTE: The DeepStack January 2022 release does not find any objects for some computers/graphics cards using GPU mode. This problem is not an On Guard issue. This issue has been reported to the DeepStack developers. Hopefully, it will be fixed soon. You may need to use the September 2021 release if you want to use the GPU mode. Also, note that performance should be improved substantially using 2022 DeepStack versions using the Threads option (see Application Settings). So, if the January 2022 DeepStack version works for you, set the Threads option and use it.

On Guard Version 3.0.1

This application works with almost any IP/network camera and adds artificial intelligence designed to identify security related pictures. More specifically, it uses DeepStack as an AI to intelligently identify objects in still (.jpg) pictures. It allows the user to monitor as many cameras as desired. It allows the user to visually create multiple areas that are of security interest, including the types of objects that are of interest in each area.

Version 3.0.1 is a significant update to previous versions. The most dramatic difference is that it fully supports any known camera that supports ONVIF or HTTP. On Guard can now obtain and analyze pictures from (1) the camera itself without disk storage (2) messages sent by a camera using FTP, and taking advantage of motion sensing capabilities of the camera (3) Blue Iris or other applications that write motion related still images to disk.

On Guard 3.0 features

1. It is provided free of charge.
2. ONVIF and HTTP based cameras are fully supported. This should include almost any camera not locked into a manufacturer proprietary system.
3. Is fully compatible with, but not reliant on Blue Iris.
4. Supports multiple cameras.
5. Support for cameras with different resolution and cameras from different manufacturers.
6. You can design areas of interest (zones) to identify or ignore activity in an area, including overlapping areas.
7. You can design your areas to be in almost any shape desired (a 64x64 grid on the pictures).
8. You can choose to base security notifications based on the size of identified objects (how big and how close they are to the camera).
9. Supports facial recognition.
10. Optionally manage the start/stop/reset of the DeepStack AI.
11. Notify any application that can take an action as a result of an HTTP message/request.
12. Supports IFTTT activation of lights and other smart devices.
13. Supports MQTT.
14. Supports email notification of clients with attached pictures. This also includes the ability to send MMS/SMS messages to phones that support it.
15. Supports Pan/Tilt/Zoom, and move to preset for cameras that support these features.
16. Tracks pictures with identified items of interest in a database so that you can move only through those pictures if desired.
17. Provides a timeline scroll bar. You can now use a track bar to scroll to pictures at any desired date/time.
18. Cleanup old/uninteresting pictures for one or all cameras.
19. Detailed documentation/instructions are included.
20. Can display snapshot pictures/video as an aid in defining areas of interest and camera positioning.

Features Not Supported

1. Does NOT store video to disk.
2. Does NOT playback videos.
3. While you can play live video (motion .jpg) directly from a camera, this capability is limited (and not smooth).
4. Does not support many doorbell cameras, and cameras locked into a proprietary camera ecosystem.

On Guard Security Assistant Version 2.0.3

This project uses Blue Iris or any IP/network camera capable of producing .jpg still images as a security camera image source. It uses DeepStack as an AI to intelligently identify objects. It allows the user to monitor as many cameras as desired. It allows the user to visually create multiple areas that are of security interest, including the types of objects that are of interest in each area. It optionally triggers recording via Blue Iris and/or sends notification emails to the user. It supports notification via MQTT with user defined (or predefined) topics and payloads.

Note that while On Guard can be used with almost any IP/network camera cameras even without Blue Iris, some functionality will be lost. For instance, you will not be able to use the pan/tilt/zoom or the Snapshot and Continuous video features. There are too many variations in the camera command layout between manufacturers and models. However, future releases are expected to support PTZ for ONVIF cameras.

Changes for 2.0.3:
Added additional Trace logging to debug email output.
Email is now sent sequentially but asynchronously to avoid overloading smtp output servers.
There is now a user configurable size limit for all picture attachments. This should automatically avoid email sender and receiver limits on maximum email sizes.

On Guard Security Assistant Version 2.0.2

This project uses Blue Iris or any IP/network camera capable of producing .jpg still images as a security camera image source. It uses DeepStack as an AI to intelligently identify objects. It allows the user to monitor as many cameras as desired. It allows the user to visually create multiple areas that are of security interest, including the types of objects that are of interest in each area. It optionally triggers recording via Blue Iris and/or sends notification emails to the user. It supports notification via MQTT with user defined (or predefined) topics and payloads.

Note that while On Guard can be used with almost any IP/network camera cameras even without Blue Iris, some functionality will be lost. For instance, you will not be able to use the pan/tilt/zoom or the Snapshot and Continuous video features. There are too many variations in the camera command layout between manufacturers and models. However, future releases are expected to support PTZ for ONVIF cameras.

This release adds optional HTML formatting for email notifications that displays attached pictures in the body of the email so that (for most email clients) it is no longer necessary to open attached pictures to view them. Because some email clients can't support this (some MMS clients in particular) there is an option to turn this on/off for each email recipient. The default is to use the new in-body email format. If this doesn't work for your email client just turn it off via clearing a checkbox available when editing the recipient.

On Guard Security Assistant Version 2.0.1

This project uses Blue Iris or any IP/network camera capable of producing .jpg still images as a security camera image source. It uses DeepStack as an AI to intelligently identify objects. It allows the user to monitor as many cameras as desired. It allows the user to visually create multiple areas that are of security interest, including the types of objects that are of interest in each area. It optionally triggers recording via Blue Iris and/or sends notification emails to the user. It supports notification via MQTT with user defined (or predefined) topics and payloads.

Note that while On Guard can be used with almost any IP/network camera cameras even without Blue Iris, some functionality will be lost. For instance, you will not be able to use the pan/tilt/zoom or the Snapshot and Continuous video features. There are too many variations in the camera command layout between manufacturers and models. However, future releases are expected to support PTZ for ONVIF cameras.

This is the first general release of Version 2. This version uses some newer Microsoft technologies (.NET Core 5). While there are a few new features this is not the primary reason for this release. Rather, it is an update in software platform, that will be the basis for all new feature releases.

This version fixes an issue in the Alpha/Pre-Release that was caused by an incompatibility between the SQL database connection string and the .NET Core libraries. The database connection string will be modified automatically to remove the reference to asynchronous processing.

Note that this version optionally supports using an XML for storing the application settings in addition to the previous Windows Registry mode. If desired you can move the camera settings to XML by starting the application in the standard Registry (non-XML mode), changing the settings mode to XML, and re-saving the camera settings. While this can be done for most other application settings it cannot be done for the AI location settings. Those would need to be re-entered manually. Using only XML for the application settings is expected to be the long term direction.

On Guard Security Assistant Version 2.0.1 (Alpha - Pre Release)

This project uses Blue Iris or any IP/network camera capable of producing .jpg still images as a security camera image source. It uses DeepStack as an AI to intelligently identify objects. It allows the user to monitor as many cameras as desired. It allows the user to visually create multiple areas that are of security interest, including the types of objects that are of interest in each area. It optionally triggers recording via Blue Iris and/or sends notification emails to the user. It supports notification via MQTT with user defined (or predefined) topics and payloads.

Note that while On Guard can be used with almost any IP/network camera cameras without Blue Iris, some functionality will be lost. For instance, you will not be able to use the pan/tilt/zoom or the Snapshot and Continuous video features. There are too many variations in the camera command layout between manufacturers and models.

This release is a new code base using the .NET Core Microsoft utility library. It requires that you download that library unless you have already done so (usually you have not).

This release adds the ability to be notified via email and/or MQTT if On Guard notices that your DeepStack AI(s) have died. In addition, the "Log Detailed Information" setting is persisted (it turns itself on if it was on the last time you used On Guard). Finally, it adds the option to store your setup information an an Xml file rather than the Windows Registry. You can turn this on and off in the Tools menu.

On Guard Security Assistant Version 1.8.2.3

This project uses Blue Iris or any IP/network camera capable of producing .jpg still images as a security camera image source. It uses DeepStack as an AI to intelligently identify objects. It allows the user to monitor as many cameras as desired. It allows the user to visually create multiple areas that are of security interest, including the types of objects that are of interest in each area. It optionally triggers recording via Blue Iris and/or sends notification emails to the user. It supports notification via MQTT with user defined (or predefined) topics and payloads.

Note that while On Guard can be used with almost any IP/network camera cameras without Blue Iris, some functionality will be lost. For instance, you will not be able to use the pan/tilt/zoom or the Snapshot and Continuous video features. There are too many variations in the camera command layout between manufacturers and models.

This release contains a bug fix that adds error detection when an AI location data is corrupt/not as expected.

Note that this is expected to be the last 1.8.x release -- new releases will be in the 2.0.x branch.

On Guard Security Assistant Version 1.8.2.2

This project uses Blue Iris or any IP/network camera capable of producing .jpg still images as a security camera image source. It uses DeepStack as an AI to intelligently identify objects. It allows the user to monitor as many cameras as desired. It allows the user to visually create multiple areas that are of security interest, including the types of objects that are of interest in each area. It optionally triggers recording via Blue Iris and/or sends notification emails to the user. It supports notification via MQTT with user defined (or predefined) topics and payloads.

Note that while On Guard can be used with almost any IP/network camera cameras without Blue Iris, some functionality will be lost. For instance, you will not be able to use the pan/tilt/zoom or the Snapshot and Continuous video features. There are too many variations in the camera command layout between manufacturers and models.

This release contains:

1. Numerous bug fixes.
2. Adds ability to turn off parked car/car overlap features, vehicle confidence increase for multiple definitions of vehicles
3. Failover for down AI servers when using multiple servers.