Issue in cnn_use_pb_tensorRT.py

[pedghz]

After running the cnn_freeze script and getting access to the /tmp/frozen_model forder , I want to run the cnn_use_pb_tensorRT.py script but I got this error message: what is going wrong here?

[TensorRT] ERROR: UFFParser: Validator error: test_model/model/decoder/upsample/unpool3/inv-res-3/inverted_residual/conv/out/LeakyRelu: Unsupported operation _LeakyRelu [TensorRT] ERROR: Failed to parse UFF model stream File "/usr/lib/python2.7/dist-packages/tensorrt/legacy/utils/init.py", line 255, in uff_to_trt_engine assert(parser.parse_buffer(stream, 0, network, model_datatype)) Traceback (most recent call last): File "/home/pedram/Desktop/bonnet-master/cnn_use_pb_tensorRT.py", line 251, in DATA_TYPE) # .HALF for fp16 in jetson! File "/usr/lib/python2.7/dist-packages/tensorrt/legacy/utils/init.py", line 263, in uff_to_trt_engine raise AssertionError('UFF parsing failed on line {} in statement {}'.format(line, text)) AssertionError: UFF parsing failed on line 255 in statement assert(parser.parse_buffer(stream, 0, network, model_datatype))

Any ideas what is going wrong here?

P.S: I am using: Ubuntu 16.04 GPU: Nvidia 1050ti Nvidia driver version: 384.130 Cuda: 9.0 Cudnn: 7 Python: 2.7 Tensroflow version: 1.13.0rc TensorRT version: 5.0.2.6

[tano297]

According to this link, under your configuration, you should probably work with cuda 10.0.130 and TF 1.12.0-rc2. Unfortunately, as a single developer, I can't follow all the spanning trees of possibilities of versions, that's why the docker image is provided. I would downgrade to TF 1.12 and then try again! Let me know if this works

[tano297]

Consider that tensorRT development is always steps behind tensorflow in terms of support. So if the gain in speed is needed, some operator/compatibility support loss is to be expected. I'm still expecting quite a few implementations in tensorRT to be able to implement more complicated models.