19
English
4. Intelligent image analysis (object recognition)
4.1. Notes on object recognition
Intelligent object detection allows you to differentiate between different types of objects (human, animal,
car) during an object detection. This gives you the option, for example, of not triggering objects at all or
triggering them with a lower priority.
Please note that object detection depends on distance, viewing angle and object size (seen from camera
to object). Please read the section "4.2 Detection range of the camera".
Note
Intelligent object detection is used to significantly reduce false alarms compared to
traditional methods of motion detection.
Traditional methods are, for example, motion
detection via heat radiation by PIR sensors or classic software motion detection without
object detection, which only takes into account the object size but not the object type.
It always requires environment-dependent individual fine-tuning of sensitivity and the use
of blanked areas to achieve maximum accuracy, but 100% accuracy and the complete
exclusion of false alarms is not possible.
Vehicles that are in the parking position do not trigger recordings until the vehicle is moving again or
there is movement in the image.
In the case of animal and person detection, a recording is triggered even if they are not moving. If a
mannequin or a picture with people is visible in the recording area, for example, please hide this area.
Note
When recognizing animals, the algorithm checks for the presence of 4 legs. The main
target group of animals to be recognized are dogs and cats.
If several objects are detected in the image at the same time, the object detected first is always displayed
in the event list.
For example, if a vehicle enters the driveway, the event list will show the vehicle as the trigger, even if
a person leaves the car in the video sequence.
The person is nevertheless marked with a colored recognition frame in the video.
Please note that object recognition searches for specific characteristics of an object in the image, but
these can be falsified by environmental influences such as wind, rain, snow, or insects. Therefore, there
is always a residual risk that an object will be wrongly classified. For example, in a person recognition
system, objects that look similar to the outline of a person can also be detected as a person. An example
could be a flag that is moving in the wind and in this case could be recognized as a person in a
mackintosh.
Summary of Contents for PPIC42520
Page 8: ...8 English 2 Hardware description 2 1 Wi Fi Pan Tilt Outdoor Camera ...
Page 9: ...9 English ...
Page 10: ...10 English 2 2 Wi Fi Outdoor Camera ...
Page 11: ...11 English ...
