Vision SDK: TIDL OD ObjectDrawLink Access to RGB values of pixels in a detection

I will check with TIDL team & get back

Hi,

To draw the bounding box around the detected object, we are using coordinates (

@param xPos
Location of the detected object in pixels along X direction.

@param yPos
Location of the detected object in pixels along Y direction.

@param objWidth
Width of the located object in pixels. Does not indicate actual
width of the object.

@param objHeight
Width of the located object in pixels. Does not indicate actual
height of the object.

)

So using these coordinates you can analyze the RGB values in the buffer & make sure the data format is in RGB or YUV

Thanks

Gaviraju

Hi,

Thanks for the information, but I am already aware what is provided with pObjectDesc type.

I am interested in accessing the buffer from inside the file objectDrawLink_algPlugin.c.
Could you please tell me which variable is the buffer, and how R,G,B values are identified with buffer, to start with ?
Or redirect me to a guide which explains it.

Thanks.

Hi,

pObjectDrawObj->drawBufAddr[0] = (UInt32)pVideoBuffer->bufAddr[0];
pObjectDrawObj->drawBufAddr[1] = (UInt32)pVideoBuffer->bufAddr[1];

This is the buffer address you can use "drawBufAddr" or pVideoBuffer->bufAddr[0] variable to access the data.

The data format seems in YUV, not in RGB. So you may need to convert from YUV to RGB.

Thanks

Gaviraju

Hi,
Thanks, that's great help.
Could you also redirect me to other files/examples where pixels (Y,U,V / R,G,B) are extracted/analyzed with drawBufAddr?
I have investigated a few examples such as AlgorithmLink_copyTrafficSign(..), but I can't figure out what to do in my particular case:

To start with, I have the following based on AlgorithmLink_copyTrafficSign(..). Could help me in tailoring this, as I am still a beginner with TIDL Buffer manipulation?

Int32 AlgorithmLink_detectRedOrGreenLight(
    UInt8 *bufAddrY,
    UInt8 *bufAddrC,
    UInt32 pitchY,
    UInt32 pitchC,
    UInt32 bufWidth,
    UInt32 bufHeight,
    UInt32 startX,
    UInt32 startY
    )
{
    UInt32 copyWidth, copyHeight;

    / align to multiple of 2, since data format is YUV420 /
    startX = SystemUtils_floor(startX, 2);
    startY = SystemUtils_floor(startY, 2);

    / clip the copy area to limit within buffer area /
    copyWidth = bufWidth;
    copyHeight = bufHeight;

    if(startX > bufWidth || startY > bufHeight)
    {
        / Nothing to copy in this case /
        return 0;
    }

    if(startX+copyWidth > bufWidth)
    {
        copyWidth = bufWidth - startX;
    }

    if(startY+copyHeight > bufHeight)
    {
        copyHeight = bufHeight - startY;
    }

    / adjust input buffer pointer to start location /
    bufAddrY = bufAddrY + pitchY*startY + startX;
    bufAddrC = bufAddrC + pitchC*startY/2 + startX;

    return 0;
}

What does bufAddrY, bufAddrC, pitchY, pitchC mean, and how do I access Y,U,V values?
After extracting Y,U,V of a pixel , I'm thinking I can then use:

void RGBfromYUV(double& R, double& G, double& B, double Y, double U, double V)
{
  R = 1.164 * Y             + 1.596 * V;
  G = 1.164 * Y - 0.392 * U - 0.813 * V;
  B = 1.164 * Y + 2.017 * U;
}

Thanks.