Google Coral Demo on Camect Home

Camect Home now compatible with the Google Coral USB accelerator. The Coral USB accelerator adds an extra Edge TPU coprocessor to the Camect inferencing (determining meaning from images) . Camect’s inferencing is intense as it detects over 30 types of objects over multiple video streams. Adding the accelerator should improve the ability to handle simultaneous detections on multiple cameras and handling busy scenes. 

To demo the accelerator we started with a brand new Coral from the box. We plugged the Coral into one of the Blue usb 3.0 ports on Camect Home, and restarted the Camect. We could tell Coral is running from the LED accelerator light. Then, we tried detecting multiple objects simultaneously from two 4K cameras. We used different people walking across the cameras to set off the detections. Since we have been using Camect for awhile, it was a noticeable improvement as the detections were a bit quicker as we looked at the Camect detection screens.

Simultaneous AI Detections using computer vision algorithm from Camect

We also have some feedback from our Camect user forum where they noticed similar improvements to our demo:

After a day of use I can confirm I do receive a lot more events than I did before the Coral USB.  For example when we arrive home with the dog most times it would just say car detected.  Now when we get out it states car, person, dog detected and also events from the overlapping camera feed from near by.  Not bad for just $50 upgrade.

Camect User Forum

Camect home by itself can handle multiple detections across many cameras at the same time as well as handle busy scenes. But adding the Google Coral USB accelerator did improve performance a bit, it was easy to install, and helped make the alerts a bit more responsive. Adding the accelerator should make things a bit faster for people who have several cameras observing busy scenes, or people who are operating close to specified limit of camera capacity on Camect Home. Adding a Coral USB stick will likely increase the number of things that are detected when the device is under high load. 

But feel free to try it out for yourself. We would love to hear your experiences

Close Menu