Analysising the motion of object from camera

Help
#1

Question/Issue:
I deploy FOMO to Nicla Vision by following the tutorial. This project can successfully detect the object and its location now. I wonder whether I can analyse the motion of the object from camera or not from edge impulse? By the way, what does the x, y, w, h in the serial port stand for (as shown in the picture.

object and motion example 1
object and motion example 11280×765 142 KB

)
Project ID:
191237
Context/Use case:
Detect the box from Nical Vision camera and its location. What to go further to analyse the motion.
ed04e22037df5023f5dd84a8b107619
ed04e22037df5023f5dd84a8b1076191283×655 131 KB

Queries about edge impulse
#2

I add a simple if else part to tell the direction of motion

#3

Hi @ryan23,

By default, FOMO only gives you the location (x, y) of the bounding box (see this talk to learn more about FOMO: tinyML Talks: Constrained Object Detection on Microcontrollers with FOMO - YouTube). I’m not sure where width and height are coming from–I’ll look into that. However, these bounding boxes are for static images only. You might be able to do some math to track an object and calculate its motion, but that’s not something Edge Impulse does out of the box.

#4

Hi @ryan23,

The width, height measurements indicate adjacent cells with the same label.

#5

Thanks for replying. I want to use the location to calculate the speed of an object. What is the unit of x and y and what do they stand for ? I am a bit confused right now

#6

Hi @ryan23,

x and y are the top-left coordinates of the bounding box, given in pixels. Here’s a reference showing the output struct of object detection: ei_impulse_result_bounding_box_t - Edge Impulse API. Hope that helps!

#7

Thanks for your help. Have a nice day

1 Like