Hi,
the mxnet and tensorrt versions of the model work like charm when that's the only thing running in the python application.When I try to run inference in a multithreading thread or in a spawned process, the face detection bbox temporarily transforms to an unkown scale when the faces are moving.
Any idea why this is happening?
Hi, the mxnet and tensorrt versions of the model work like charm when that's the only thing running in the python application.When I try to run inference in a multithreading thread or in a spawned process, the face detection bbox temporarily transforms to an unkown scale when the faces are moving. Any idea why this is happening?