Hi,
I am trying to load NVidia Peoplenet with TensorRT 7.1 on a tesla V100 using python.
I copied tlt-converter from the container to the host and got the saved.engine file. But how can I load the engine to perform inference using python?
I didn’t find any tutorial for this purpose.
Thanks in advance
1 Like
Pidem
2
Facing the same issue: I tried using this code, but keep getting errors:
inference.txt (2.5 KB)
[TensorRT] ERROR: Parameter check failed at: engine.cpp::setBindingDimensions::949, condition: profileMinDims.d[i] <= dimensions.d[i]
[TensorRT] WARNING: Current optimization profile is: 0. Please ensure there are no enqueued operations pending in this context prior to switching profiles
[TensorRT] ERROR: Parameter check failed at: engine.cpp::setBindingDimensions::949, condition: profileMinDims.d[i] <= dimensions.d[i]
test
[TensorRT] ERROR: Parameter check failed at: engine.cpp::enqueueV2::460, condition: !mEngine.hasImplicitBatchDimension()
I did specify the batch size and the dimension size during the tlt-converter step.
Pidem
3
I was able to get the detections, but I have no idea how to parse the results in python it would be really helpful !
inference.txt (4.5 KB)