segmentation fault when run mContext->execute(batch_size, inferenceBuffers)

I make a plugin for the detector follow the example in vat-nvidia/deepstrean-plugins. My detector use tensorrt to do inference.
When I run nvgstiva-app -c path_to_config, it will meet segmentation fault when execute
mContext->execute(batch_size, inferenceBuffers) in the second times. In the first time, mContext->execute can be executed. how could I solve this problem?


Could you try some simple pipeline with your plugin to make sure it works correctly first?

wayne zhu

Hi 373197201,

Have you tried with our suggestion?
Any status could be shared or issue have been resolved?