3.Basic settings

Indicates the basic operations for setting the AI Scene Change Detection function.

Each item has a bubble mark. Click to display the procedure in the lower left corner of the screen.
Refer to the following for precautions on setting the angle of view.
1.3 Installation conditions of camera
The [AI Scene change detection] function is implemented with the following elements.
The following explains the terms used in the following explanations.
Learning images
Still images used for learning. Images taken directly from the camera are mainly used. They are saved on the SD Memory Card.
Detector
This is the result of learning scene change detection created by AI from multiple learning images. It is saved on the SD Memory Card.
Detection area
The area to be covered by scene change detection. Specified by a blue frame on the live image.
Configure the Detector.

There are three areas in which Detector can be recorded: when you want to switch models in black-and-white video, or when you want to try it in several patterns.
For example, by having Detector 1 study a color image and Detector 2 study a black-and-white image, scene change detection can be performed using Detector 1 for a color image and Detector 2 for a black-and-white image. However, when Detector is switched in black-and-white video, specify detector number in the advanced [Detector used in black and white video].
4.3.1 Advanced detection settings
Select [Detector number].
-
Select from 1 to 3.
-
[(Used)] is displayed for the number where a Detector is already used.
-
A number that is not in use will be marked [(Unused)].
Set [Detector Name].
-
Set the name of the detector selected in [Detector number]. Up to 20 half-width or full-width characters can be set.
Press [Set].
Note that if you select a [(Used)] model number, the Detector will be overwritten at the completion of learning.
Set the detection area.
1.
Draw the detection area in the live image on the iCT individual setting screen.
-
The detection area can be set as a rectangle of any shape.
2.
Select the detection area number to be set.
-
Select 1:(Detection area name) ~ 16:(Detection area name) from [All].
-
Select [All] to set each detection area individually, or to set all detection areas collectively.
When [All] is selected, the checkbox will be added next to the parameters as shown.
Selecting a check mark on the item to be set collectively will apply the selected value.
[Detection area group name] cannot be set when [All] is selected.
3.
Set the [Detection area group name]
-
You can set up to 20 characters of both half-width and full-width characters.
4.
Select [Detection time]
-
Select the time from when an error is detected to when an alarm is sent.
-
Select [2s], [3s], [5s], [10s], [20s], [30s], [40s], [50s], [1min], [5min], [10min], [15min], [30min], or [1h].

・[Image change threshold] and [Area change threshold] are adjusted when checking the accuracy.


See below for details.
3.5 Confirm the learning results

・If [Detection time] is shorter than [Dwell time] of [Day & Night(IR)] of camera setting, AI scene change detection alarm may be activated at the time when an image is switched to black-and-white. It can be improved by setting [Dwell time] of [Day & Night(IR)] shorter than [Detection time].


5.
Select [State inversion]
-
When [On] is selected, the alarm will be activated in the same way as when learning. When [Off] is selected, the alarm will be activated in a way that is different from the one learned.
6.
Select [Abandonment detection]
Select [On] for detecting obsolete objects (suspicious objects, etc.).

・When [Abandonment detection] is set to [On], objects of a certain size are detected regardless of the [Area change threshold] setting. (Reference: Approx. 5 to 50% width of screen)


・If you set [Abandonment detection] to [On] for more than one [Detection area], then [Image change threshold] and [Detection time] are the lowest numbered values for [Detection area].


・Even when [Exclude a person from scene change] is set to [On], shadows may be detected as abandoned objects. In that case, increasing the [Image change threshold] may suppress false detection.


・In the abandoned position detection, set the detection area larger so that the center of the bottom surface of the object is within the detection area.


Image of setting the detection area

7.
Press [Set].
Note that the setting is cleared by switching the detection area number before pressing the set button or by pressing the [Back] or [Next] button.
Images are taken for learning.

You can take pictures directly from installed cameras and select from [Auto save] and [Manual save].
In [Auto save], when [Save interval] and [Number of save images] are specified, one still image for [Number of save images] is automatically saved for each [Save interval].
In [Manual save], you can manually save still images when necessary.
If the image is not saved properly when pressing [Save], restart the camera.
1.
Select [Saving learning images]
-
Select [Manual save].
2.
Press [Save]
-
One shot image (still image) is capture.
1.
Select [Saving learning images]
-
Select [Auto save].
2.
Set the [Save interval]
-
Select from [1min], [5min], [10min], [15min], [30min], or [60min].
3.
Select [Number of save images]
-
Select from [10], [20],..., [90], [100], [150], or [200].
4.
Press the [Start]
-
To stop auto save, press [Stop].
-
During auto save, progress is displayed beside the Stop button.
Perform learning.

Press the Display button to open the Learning images selection window.
Select the image to be used for learning from the thumbnail list of captured images.
1.
Select the images (Training images) to be used for learning.
-
Select the checkbox in the lower left corner of thumbnail of the target image as shown in the figure.
-
Learning requires at least two images. Up to 200 images can also be learned.
2.
Press [Start]
Check the learning accuracy.

Press [Display] to start the demo screen.
Refer to the following for details about the demo screen.
5. Demo screen
Press [Back] three times to return to [Detection area settings].
[Image change threshold] and [Area change threshold] are adjusted so that various possible situations are reproduced and alarms are not generated when the alarm does not want to be detected.
1.
Adjust the [Image change threshold]
-
Set scene change thresholds.
The smaller the value, the easier it is to detect the change. The larger the value, the more difficult it is to detect the change.
The area where the change is detected is displayed in red on the demo screen.
2.
Adjust the [Area change threshold]
-
Set the area of the scene change.
The alarm is activated when the set detection time has elapsed while the scene change Area change ratio in the detection area exceeds this threshold.
The Area change ratio is displayed on the demo screen.