Apologies for the incorrect answer above.
The outputs you get from the IMX500 at the application level should already be de-quantized (this happens in libcamera). So the quantized values from the IMX500 are converted to fp32 values AFTER proper shift and scale. If you suspect this is not what is happening, please do share more information or preferably a code sample so we can investigate further.
The outputs you get from the IMX500 at the application level should already be de-quantized (this happens in libcamera). So the quantized values from the IMX500 are converted to fp32 values AFTER proper shift and scale. If you suspect this is not what is happening, please do share more information or preferably a code sample so we can investigate further.
Statistics: Posted by naushir — Mon Mar 17, 2025 8:53 am