[onnx->TensorRT]Parser problem about padding

lulindeng073 · April 19, 2021, 2:34am

Hi,

I’m trying to accelerate model inference speed by TensorRT, the model has been first convert to onnx format from tensorflow saved model using tf2onnx .
When I parse the onnx model using tensorrt.OnnxParser(), I got this error:

[TensorRT] VERBOSE: ModelImporter.cpp:103: Parsing node: generator/G_MODEL/A/MirrorPad [Pad]
[TensorRT] VERBOSE: ModelImporter.cpp:119: Searching for input: generator_input:0
[TensorRT] VERBOSE: ModelImporter.cpp:119: Searching for input: const_fold_opt__1670
[TensorRT] VERBOSE: ModelImporter.cpp:125: generator/G_MODEL/A/MirrorPad [Pad] inputs: [generator_input:0 -> (1, -1, -1, 3)], [const_fold_opt__1670 -> (8)], 
ERROR: Failed to parse the ONNX file.
number of errors: 1
In node -1 (importPad): UNSUPPORTED_NODE: Assertion failed: mode == "constant" && value == 0.f && "This version of TensorRT only supports constant 0 padding!"

I read some related issues posted by others, It seems like TensorRT can only support constant 0 padding, but in my model, the pad in reflect mode. In this case, what can I do to solve this problem?

Deepest thanks for your reply!

Environment

TensorRT Version: 7.2.3.4
GPU Type: GTX 1080
Nvidia Driver Version: 450.51.06
CUDA Version: 10.0
CUDNN Version: 7.6.4
Operating System + Version: Ubuntu 16.04
Python Version (if applicable): 3.7.10
TensorFlow Version (if applicable): tensorflow-gpu 1.15.0
PyTorch Version (if applicable):
Baremetal or Container (if container which image + tag):

NVES · April 19, 2021, 2:37am

Hi,
Request you to share the ONNX model and the script if not shared already so that we can assist you better.
Alongside you can try few things:

validating your model with the below snippet

check_model.py

import sys
import onnx
filename = yourONNXmodel
model = onnx.load(filename)
onnx.checker.check_model(model).
2) Try running your model with trtexec command.
https://github.com/NVIDIA/TensorRT/tree/master/samples/opensource/trtexec
In case you are still facing issue, request you to share the trtexec “”–verbose"" log for further debugging
Thanks!

lulindeng073 · April 19, 2021, 3:33am

Thanks for you prompt reply!

I run the check_model.py with my onnx model, it has no warning or errors.
I build the trtexec and try running my model with trtexec command:

&&&& RUNNING TensorRT.trtexec # /home/lulin/TensorRT/TensorRT-7.2.3.4/bin/trtexec --onnx=hayao_op11.onnx --verbose
[04/19/2021-11:32:29] [I] === Model Options ===
[04/19/2021-11:32:29] [I] Format: ONNX
[04/19/2021-11:32:29] [I] Model: hayao_op11.onnx
[04/19/2021-11:32:29] [I] Output:
[04/19/2021-11:32:29] [I] === Build Options ===
[04/19/2021-11:32:29] [I] Max batch: explicit
[04/19/2021-11:32:29] [I] Workspace: 16 MiB
[04/19/2021-11:32:29] [I] minTiming: 1
[04/19/2021-11:32:29] [I] avgTiming: 8
[04/19/2021-11:32:29] [I] Precision: FP32
[04/19/2021-11:32:29] [I] Calibration: 
[04/19/2021-11:32:29] [I] Refit: Disabled
[04/19/2021-11:32:29] [I] Safe mode: Disabled
[04/19/2021-11:32:29] [I] Save engine: 
[04/19/2021-11:32:29] [I] Load engine: 
[04/19/2021-11:32:29] [I] Builder Cache: Enabled
[04/19/2021-11:32:29] [I] NVTX verbosity: 0
[04/19/2021-11:32:29] [I] Tactic sources: Using default tactic sources
[04/19/2021-11:32:29] [I] Input(s)s format: fp32:CHW
[04/19/2021-11:32:29] [I] Output(s)s format: fp32:CHW
[04/19/2021-11:32:29] [I] Input build shapes: model
[04/19/2021-11:32:29] [I] Input calibration shapes: model
[04/19/2021-11:32:29] [I] === System Options ===
[04/19/2021-11:32:29] [I] Device: 0
[04/19/2021-11:32:29] [I] DLACore: 
[04/19/2021-11:32:29] [I] Plugins:
[04/19/2021-11:32:29] [I] === Inference Options ===
[04/19/2021-11:32:29] [I] Batch: Explicit
[04/19/2021-11:32:29] [I] Input inference shapes: model
[04/19/2021-11:32:29] [I] Iterations: 10
[04/19/2021-11:32:29] [I] Duration: 3s (+ 200ms warm up)
[04/19/2021-11:32:29] [I] Sleep time: 0ms
[04/19/2021-11:32:29] [I] Streams: 1
[04/19/2021-11:32:29] [I] ExposeDMA: Disabled
[04/19/2021-11:32:29] [I] Data transfers: Enabled
[04/19/2021-11:32:29] [I] Spin-wait: Disabled
[04/19/2021-11:32:29] [I] Multithreading: Disabled
[04/19/2021-11:32:29] [I] CUDA Graph: Disabled
[04/19/2021-11:32:29] [I] Separate profiling: Disabled
[04/19/2021-11:32:29] [I] Skip inference: Disabled
[04/19/2021-11:32:29] [I] Inputs:
[04/19/2021-11:32:29] [I] === Reporting Options ===
[04/19/2021-11:32:29] [I] Verbose: Enabled
[04/19/2021-11:32:29] [I] Averages: 10 inferences
[04/19/2021-11:32:29] [I] Percentile: 99
[04/19/2021-11:32:29] [I] Dump refittable layers:Disabled
[04/19/2021-11:32:29] [I] Dump output: Disabled
[04/19/2021-11:32:29] [I] Profile: Disabled
[04/19/2021-11:32:29] [I] Export timing to JSON file: 
[04/19/2021-11:32:29] [I] Export output to JSON file: 
[04/19/2021-11:32:29] [I] Export profile to JSON file: 
[04/19/2021-11:32:29] [I] 
[04/19/2021-11:32:29] [I] === Device Information ===
[04/19/2021-11:32:29] [I] Selected Device: GeForce GTX 1080
[04/19/2021-11:32:29] [I] Compute Capability: 6.1
[04/19/2021-11:32:29] [I] SMs: 20
[04/19/2021-11:32:29] [I] Compute Clock Rate: 1.873 GHz
[04/19/2021-11:32:29] [I] Device Global Memory: 8116 MiB
[04/19/2021-11:32:29] [I] Shared Memory per SM: 96 KiB
[04/19/2021-11:32:29] [I] Memory Bus Width: 256 bits (ECC disabled)
[04/19/2021-11:32:29] [I] Memory Clock Rate: 5.005 GHz
[04/19/2021-11:32:29] [I] 
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::GridAnchor_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::NMS_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::Reorg_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::Region_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::Clip_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::LReLU_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::PriorBox_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::Normalize_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::RPROI_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::BatchedNMS_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::BatchedNMSDynamic_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::FlattenConcat_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::CropAndResize version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::DetectionLayer_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::Proposal version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::ProposalLayer_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::PyramidROIAlign_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::ResizeNearest_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::Split version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::SpecialSlice_TRT version 1
[04/19/2021-11:32:29] [V] [TRT] Registered plugin creator - ::InstanceNormalization_TRT version 1
----------------------------------------------------------------
Input filename:   hayao_op11.onnx
ONNX IR version:  0.0.6
Opset version:    11
Producer name:    tf2onnx
Producer version: 1.8.4
Domain:           
Model version:    0
Doc string:       
----------------------------------------------------------------
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::GridAnchor_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::NMS_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::Reorg_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::Region_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::Clip_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::LReLU_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::PriorBox_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::Normalize_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::RPROI_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::BatchedNMS_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::BatchedNMSDynamic_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::FlattenConcat_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::CropAndResize version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::DetectionLayer_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::Proposal version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::ProposalLayer_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::PyramidROIAlign_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::ResizeNearest_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::Split version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::SpecialSlice_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] Plugin creator already registered - ::InstanceNormalization_TRT version 1
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:202: Adding network input: generator_input:0 with dtype: float32, dimensions: (1, -1, -1, 3)
[04/19/2021-11:32:30] [V] [TRT] ImporterContext.hpp:120: Registering tensor: generator_input:0 for ONNX tensor: generator_input:0
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/out_layer/Conv/weights/read__84
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/E/LayerNorm_2/gamma/read__83
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/E/LayerNorm_1/gamma/read__80
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/E/LayerNorm/gamma/read__77
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/E/Conv_2/weights/read__81
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/E/Conv_1/weights/read__78
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/E/Conv/weights/read__75
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/D/mul/x:0
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/D/LayerNorm_1/gamma/read__74
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/D/LayerNorm/gamma/read__71
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/D/Conv_1/weights/read__72
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/D/Conv/weights/read__69
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r4/r4/bias/read__60
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r4/LayerNorm/gamma/read__58
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r4/Conv_1/weights/read__63
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r4/Conv/weights/read__56
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r4/2/gamma/read__65
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r4/1/gamma/read__62
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r3/r3/bias/read__50
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r3/LayerNorm/gamma/read__48
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r3/Conv_1/weights/read__53
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r3/Conv/weights/read__46
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r3/2/gamma/read__55
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r3/1/gamma/read__52
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r2/r2/bias/read__40
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r2/LayerNorm/gamma/read__38
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r2/Conv_1/weights/read__43
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r2/Conv/weights/read__36
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r2/2/gamma/read__45
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r2/1/gamma/read__42
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r1/r1/bias/read__30
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r1/LayerNorm/gamma/read__28
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r1/Conv_1/weights/read__33
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r1/Conv/weights/read__26
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r1/2/gamma/read__35
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/r1/1/gamma/read__32
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/LayerNorm_1/gamma/read__68
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/LayerNorm/gamma/read__25
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/Conv_1/weights/read__66
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/C/Conv/weights/read__23
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/B/LayerNorm_1/gamma/read__22
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/B/LayerNorm/gamma/read__19
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/B/Conv_1/weights/read__20
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/B/Conv/weights/read__17
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/A/LayerNorm_2/gamma/read__16
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/A/LayerNorm_1/gamma/read__13
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/A/LayerNorm/gamma/read__10
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/A/Conv_2/weights/read__14
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/A/Conv_1/weights/read__11
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: generator/G_MODEL/A/Conv/weights/read__8
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_two__403
[04/19/2021-11:32:30] [W] [TRT] onnx2trt_utils.cpp:220: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32.
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_starts__394
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1670
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1665
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1664
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1663
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1661
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1655
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1653
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1652
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1651
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1649
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1646
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1645
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1644
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1643
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1641
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1638
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1637
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1635
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1634
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1632
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1627
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1626
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1621
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1600
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1590
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1588
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1587
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1584
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1565
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1564
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1561
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_fold_opt__1557
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_ends__333
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_empty_float__347
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: const_axes__334
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:90: Importing initializer: Const__1482
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:103: Parsing node: generator/G_MODEL/A/MirrorPad [Pad]
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:119: Searching for input: generator_input:0
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:119: Searching for input: const_fold_opt__1670
[04/19/2021-11:32:30] [V] [TRT] ModelImporter.cpp:125: generator/G_MODEL/A/MirrorPad [Pad] inputs: [generator_input:0 -> (1, -1, -1, 3)], [const_fold_opt__1670 -> (8)], 
ERROR: builtin_op_importers.cpp:2261 In function importPad:
[8] Assertion failed: mode == "constant" && value == 0.f && "This version of TensorRT only supports constant 0 padding!"
[04/19/2021-11:32:30] [E] Failed to parse onnx file
[04/19/2021-11:32:30] [E] Parsing model failed
[04/19/2021-11:32:30] [E] Engine creation failed
[04/19/2021-11:32:30] [E] Engine set up failed
&&&& FAILED TensorRT.trtexec # /home/lulin/TensorRT/TensorRT-7.2.3.4/bin/trtexec --onnx=hayao_op11.onnx --verbose

I think it still facing the padding mode issue, should I need to add the custom layer?

Thanks!

spolisetty · April 19, 2021, 7:10am

Hi @lulindeng073,

TRT doesn’t support reflect padding. You may need to re-train the model with zero padding.

Thank you.

derekmagee · August 17, 2021, 4:41pm

Any chance of putting in a feature request for this? Mirror padding is getting more common as it reduces edge effects when analysing large images in tiles. e.g GitHub - LynnHo/CycleGAN-Tensorflow-2 uses it in the mapping function. I was planning to update all my UNET code to use it after seeing the results from that.

user5204 · October 19, 2021, 4:38pm

This does not seem too difficult to implement. Please consider adding. Thanks!!

Topic		Replies	Views
Trtexec cannot convert QAT onnx model to trt model TensorRT tensorrt	7	1174	January 26, 2023
Tensorrt8.2 convert .onnx model to .trt ,throw "two inputs (data and weights) are allowed only in explicit-quantization mode"error TensorRT	9	1957	April 6, 2022
Error converting gru model to TRT engine TensorRT	3	1008	October 25, 2021
Pytorch model convert to TensorRT engine failed TensorRT tensorrt , pytorch , onnx	5	1213	December 28, 2020
ONNX to TensorRT conversion Jetson Nano tensorrt , pytorch , onnx	4	360	July 3, 2024
Convert onnx to trt format using trtexec TensorRT tensorrt	5	5372	August 17, 2021
Running trtexec on Google's EfficientDet-D4 in ONNX Format TensorRT	9	1322	June 11, 2021
Run the Tacotron2 meet the problem TensorRT	7	1834	February 21, 2022
[TensorRT] ERROR: Network must have at least one output [TensorRT] ERROR: Network validation failed TensorRT tensorrt , cuda , onnx	10	2775	October 16, 2020
Parse onnx file failed: Parameter check failed. condition:condition: allDimsGtEq(windowSize, 1) && volume(windowSize) < MAX_KERNEL_DIMS_PRODUCT(nbSpat TensorRT tensorrt	3	948	May 17, 2022

[onnx->TensorRT]Parser problem about padding

Environment

check_model.py

Related topics