Get depth of moving object

gort

Hey all, pretty new to DepthAI so please bear with me, thanks in advance.

Using open CV I've managed to get my OAK-D PRO camera to reliably track and output 2D coordinates of a ping pong ball. I would like to utilize the depth feature and output 3D coordinates of the ball as well. I've been looking at the examples folder, specifically /SpatialDetection/spatial_location_calculator.py, thinking about how to manipulate that to fit my purpose.

In my understanding, there is a ROI that is the region that the depth is calculated. Would I be able to manipulate the ROI to only give me the depth of a certain coordinate per frame? Would appreciate any pointers, there's a bit too much code I don't understand so I'm asking for help.

Thanks again!

jakaskerl

gort
Check https://docs.luxonis.com/software/perception/rgb-d/ guide. You need to align depth to RGB, then you can use the same pixel indexing on both the RGB (where you use opencv) as well as depth.

To update roi on the fly: luxonis/depthai-pythonblob/main/examples/SpatialDetection/spatial_location_calculator.py#L150-L156

Thanks,
Jaka