TDA4VH-Q1: Creating an object array from multiple object arrays

Hi Rishitha,

Rishitha Gollamudi said:

I currently have separate object arrays for each camera. Is there a way to combine these different object arrays into a single array that can be sent to the SRV node? 

I suppose that should be doable with openvx.

Could you please share some code snippet for the arrays on each side?
That will help me find someone who can answer the question for you.

Hi Gang Hua,

Thank you for your response. I have attached the relevant code snippets for both sides.

In the first image, you can see that we have five separate object arrays, each corresponding to a different camera. These arrays are being assigned to an array of arrays (srv_input_frames). However, the SRV node only accepts a single object array that contains multiple items, as shown in the second image.

Could you please advise on how we can combine these five individual object arrays (each containing a single item) into one object array with five items, so that it can be passed to the SRV node?

Hi Rishitha,

Thanks for sharing the details!

Let me transfer the question to a s/w expert of the SRV demo.

Hi,

Few APIs below that could be of use to you.

vxGetObjectArrayItem() -> to get the vx_image from object array

vxQueryObjectArray() -> to get number of items in an object array

One way I see is as shown below

Now what you could do here will be to create an object array of size 4 (this would be the input of SRV node)

then extract each reference images using vxGetObjectArrayItem(), So now you will have 4 vx_images. Now you can create 4 individual object_arrays with these and pass this to source nodes

Regards,

Nikhil

Hi Nikhil,

We have a different scenario, as shown in the below image. In this setup, we have five separate graphs, each corresponding to a different camera. From each graph, we are extracting the VISS output, which is a single object array for each camera.

Additionally, we have a separate graph for the SRV, which includes an SRV node. Our objective is to merge the five object arrays from the camera graphs into a single array, containing five items—one for each camera, as shown in the image. This merged array will then be sent to the SRV node.

We would like to understand how the merging process (highlighted in red in the image) is implemented.