Creating a Count Application
  • 08 Feb 2024
  • 3 Minutes to read
  • Dark
    Light

Creating a Count Application

  • Dark
    Light

Article summary

33 STEPS

1. The first step is to open Cogniac and select Count from the Vision AI Applications. You may notice many different options, but for now we will select Count under Vision AI Applications.

Step 1 image

2. Select the Name to update the name of the Count application

Step 2 image

3. Type Name of the Application. Please keep in mind that the name should be something meaningful to the purpose of the Application. For example if you are counting animals, you can name it Animals-Counter or animals-count-application

Step 3 image

4. Once you have selected the name of the Count application click Next

Step 4 image

5. Type an Input Subjects name. Again it is good to consider a name that will be similar to the subjects of interest displayed in the image. For example in our case we are counting animals, so we named it as animals

Step 5 image

6. Click Create a new subject animals

Step 6 image

7. Click Next blue button to proceed to the Output Subjects step

Step 7 image

8. The Output Subject of the Count application is where you will find the labeled and sorted images. Usually the Output Subject gets populated with media (images/ videos) after successful Training of the Application. 

Step 8 image

9. Type an Output Subjects name. In the Count application you can have only one output subject.

Step 9 image

10. Click Create a new subject in our case we have chosen animals as we will count animals displayed in the Media.

Step 10 image

11. Click Next

Step 11 image

12. On the last screen you would see some advanced settings such as Maximum Count, Minimum Count, Frames Per Second Requirement. Click Maximum Count 

Step 12 image

13. Click Frames per Second Requirement

Step 13 image

14. Type Frames per Second Requirement in a numeric value

Step 14 image

15. Scroll down and click Complete blue button

Step 15 image

16. Select the Input Subject, which is always displayed in the beginning of the Application Pipeline. In current case select animals.

Step 16 image

17. Click Upload Media to add the Initial Dataset (a set of media - images or video which will be processed by the Application Model)

Step 17 image

18. Click select files or select folder to upload the initial Dataset of images/ videos to the Input Subjects of the Count Application

Step 18 image

19. Click Next button once you have selected the images or the folder containing the media.

Step 19 image

20. Click Upload 

Step 20 image

21. Click on the application name which in our case is the animals-count-application

Step 21 image

22. Click Replays

Step 22 image

23. Select the Input Subject. In our case this is animals

Step 23 image

24. Select the Probability filter

Step 24 image

25. With your mouse drag and drop the Probability Filter from the beginning of the scale.

Step 25 image

25b. Drop

Step 25b image

26. Scroll down and click Force Feedback

Step 26 image

27. Click Replay Subjects blue button in the end of this tab

Step 27 image

28. Select the Provide Feedback option

Step 28 image

29. Click on the + or - buttons to select the number of Subjects Of Interest shown in the image 

Step 29 image

30. Click Confirm once you have selected the proper number of the Subjects displayed in the image.

Step 30 image

31. During providing feedback there are also: BackSideline and Skip options.

Step 31 image

32. Click Quit once there are No pending feedback for the selected application.

Step 32 image

33. That's it. You're done.

Step 33 image

Here's an interactive tutorial

** Best experienced in Full Screen (click the icon in the top right corner before you begin) **

https://www.iorad.com/player/2274888/Cogniac---How-to-create-a-Count-Application-and-provide-feedback

I. What the app is doing

The Count application is used to count the number of subjects of interest in an image

II. Use case / where the app can be used in real life:

  • Manufacturing

  • Gambling business - counting dice, etc.

  • Counting money at banks / ATMs etc.

III. What could cause issues while using the app:

  • Sensitivity to image quality: The accuracy of counting apps can also be affected by the quality of the image being processed. If the image is blurry or noisy, the counting app may need help identifying and counting objects accurately.

  • Media is being captured from different angles.

  • Sensitivity to object appearance: The accuracy of counting apps can be affected by the appearance of the objects being counted. For example, if the objects are similar in appearance or are overlapping, the counting app may face difficulties counting them accurately.

  • Computational requirements: Counting apps can be computationally expensive to run, especially when counting large numbers of objects or when using complex machine-learning algorithms

  • Training data: Counting apps require training data to learn how to identify and count different types of objects. This training data can be difficult and time-consuming to collect.


Was this article helpful?