Camlytics use case: people counting

One of the most popular use cases for Camlytics camera software is counting various objects in a video stream (webcam, CCTV IP camera, DVR / NVR, files), especially people (pedestrians, store visitors, etc.). It can be a grocery store, mall, parking lot, street, you name it - people counting is useful everywhere.

Very often you need to detect only people walking through special direction - this can be done easily by setting up a tripwire with appropriate direction activated.

For best possible counting precision it's recommended to use overhead cameras that are installed on the ceiling and looking vertically down. In this case, moving objects (people) will not overlap each other in a camera view so the proper tracking separation will be possible thus improving counting results.

It is also important to keep in mind that proper calibration of Camlytics software is essential for proper object tracking, especially with an overhead camera. You can read more on calibration in the documentation. You can see the properly calibrated video scene in the image below - moving green frames are mostly the same size as the people.

Low video resolution does not affect the precision, in fact, the lower the resolution - the better performance you are going to get. See the video example of the working solution with the highest precision (overhead camera) and performance (low resolution):

Also, you can send all video people counting events as emails via single alerts or as daily aggregated reports. Developers and integrators can customize the counting events output - send them to a custom server, database, UI, etc via convenient and simple Camera Events API. In case you dont have any coding skills - don't worry, you can upload all your events automatically to the cloud with Camlytics webhooks.

Don't forget that you can run video analytics (people counting) on pre-recorded video files and folders and get huge batches of files processed unattended.

Also, check out our YouTube channel which has plenty of real-life video analysis demos.