@nvidians
hii I am trying to load frcnn engine file in a python script and wants to run predictions on test images.
I am using these bindings:
this is my complete script:
frcnn.py (8.6 KB)
but I am getting this issue:
Any help will be really appreciable.
Thanks
Morganh
February 17, 2021, 3:19pm
2
Please try Inferring detectnet_v2 .trt model in python - #44 by Morganh
TRT_LOGGER = trt.Logger(trt.Logger.WARNING)
trt.init_libnvinfer_plugins(TRT_LOGGER, ’ ')
@Morganh hii … I have added that piece of code in my script.
after running this I am facing this issue :
are these bindings wrong for frcnn:
binding_to_type = {
“input_1”: np.float32,
“dense_class_td/Softmax,dense_regress_td”: np.float32,
“BiasAdd,proposal”: np.float32,
}
any idea? what I need to change ?
thanks
Morganh
February 18, 2021, 7:40am
5
Can you ignore binding_to_type and set dtype=np.float32 directly?
@Morganh
as per your suggestion I tried like this :
but getting issue:
thanks
Morganh
February 19, 2021, 9:47am
7
Please try to debug your code by yourself.
For your latest issue, I suggest you to delete
dtype = [str(binding)]