I have two cameras placed facing each other.
Then I use Fusion for BodyTracking, and the image-related data such as keypoint_2d, boundbox_2d, and mask are blank in the obtained data.
What I want to get is: the original data corresponding to the fused data, so that I can get the boundbox_2d of the front and back of a person, and then cut out the image of the person.
However, the API does not seem to provide such a function. And I don’t want to make it myself to determine which two bone data from different cameras are the same object, because my programming ability is low and my program still has problems with positioning accuracy.
Is there any way? Or can the API be added with the function of obtaining the corresponding original data in the future?
Hi,
If I understand correctly, do you want access to one camera’s body tracking data, not the fused data?
If so, you can access it by using the camera ID (see : https://github.com/stereolabs/zed-sdk/blob/master/body%20tracking/multi-camera/cpp/src/main.cpp#L168C24-L168C38)
Stereolabs Support
Sorry, I know this api. But this is not what I want.
What I want to say is, for example:
There are three people on the field and camera A and camera B;
Camera A gets BodyData: A1, A2, A3;
Camera B gets BodyData: B1, B2, B3;
A1 and B2 are the same person,
A2 and B3 are the same person,
A3 and B1 are the same person;
And what I want is to know the correspondence above.
I see that Fusion can calculate one data from multiple cameras, so I assume that in the process it knows which data belongs to the same person. But the data obtained only contains bones.
I have tried to judge whether the data belongs to the same person directly based on the distance, but the effect is not good, so I tried Fusion.
Can you understand my intention with this example?
Ok, I see, thanks for the clarification.
This information is currently not available in the ZED SDK. I shared your feedback with the team and we’ll decide if this is a feature we can add in a future release or not.
Thanks!
Stereolabs Support
OK, thank you very much