Counting moving objects is a common task in the present. Numerous areas of human activities require counting various types of moving objects. For example, people visiting shopping malls, traffic on busy roads, products on a factory line, and many others. Previously, the hardware for doing this job was too expensive. Technology has improved and this task can be done successfully with a pocket-sized computer that won't cost you an arm and a leg.
Lines are fundamental terms in object counting. You count an object when it crosses the line. One direction is "in" and the other is "out". The position of the starting point defines directions. If you would like to reverse the directions you should just swap positions of the start and end points for that particular line.
We support an unlimited number of lines. You can also select which objects you don't want to count on a particular line.
Line moving and resizing is as easy as dragging its two terminal points.
Updating all the line settings is performed in a dialog you invoke by double-clicking on one of its points.
The new line you may create simply by dragging the mouse from the starting to the ending point of the line.
The dialog for defining a line is shown on the image on the right.
In every frame of the video input, objects are counted if their central position is inside the particular zone polygon.
As a common frame rate is 30 fps, this means that the zone occupancy would be recalculated every 30th of a second, which would be difficult to work with.
That's why for each zone there is a parameter "frames in average occupancy" which simply as a unit of reporting gives you a total count of occurrences of each class of objects in the particular zone averaged over that frames count.
So, if you have a video input frame rate equal to 30 fps, you set the frames in average occupancy to 30, you will get a zone count reported every second for that zone. 300 will result in a zone count every 10 seconds.
For every zone, you can also set up which object classes you don't want to count in it. Just add them to the list of ignored classes.
The most convenient part is that you can change the zone shape with simple mouse clicks on the points defining the zones.
To move a point, just do drag&drop to the destination.
To add a new point to the polygon, click on it while pressing the CTRL key on the keyboard, and the newly created point will be inserted between that point and the subsequent one.
To delete a point in the polygon, click on it while pressing the ALT key on the keyboard.
Classes of objects in training artificial intelligence to recognize and detect them in the image.
The classes used in this particular setting are from the road traffic application.
You can make some of them ignored by setting them as inactive.
If the detection confidences are having low values for your particular camera setting then you can try to lower the threshold value. Also if you have similar objects wrongly detected, you can raise the threshold value to filter them out.
To summarize this product presentation I can emphasize that it is highly customizable, so it is readily applicable to other fields.
Learning the network for a completely new set of classes of objects is a simple task of annotation several hundreds of images.
The hardware on which it works is readily available and cheap. It is also open source platform, ready to connect over all kinds of protocols, connecting to other devices with all kinds of interfaces.To summarize this product presentation I can emphasize that it is highly customizable, so it is readily applicable to other fields.
Learning the network for a completely new set of classes of objects is a simple task of annotation several hundreds of images.
The hardware on which it works is readily available and cheap. It is also open source platform, ready to connect over all kinds of protocols, connecting to other devices with all kinds of interfaces.