Tensorrt optimization tool dynamic inputs

Please provide the following info (tick the boxes after creating this topic):
Software Version
DRIVE OS 6.0.10.0
DRIVE OS 6.0.8.1
DRIVE OS 6.0.6
DRIVE OS 6.0.5
DRIVE OS 6.0.4 (rev. 1)
DRIVE OS 6.0.4 SDK
other

Target Operating System
Linux
QNX
other

Hardware Platform
DRIVE AGX Orin Developer Kit (940-63710-0010-300)
DRIVE AGX Orin Developer Kit (940-63710-0010-200)
DRIVE AGX Orin Developer Kit (940-63710-0010-100)
DRIVE AGX Orin Developer Kit (940-63710-0010-D00)
DRIVE AGX Orin Developer Kit (940-63710-0010-C00)
DRIVE AGX Orin Developer Kit (not sure its number)
other

SDK Manager Version
2.1.0
other

Host Machine Version
native Ubuntu Linux 20.04 Host installed with SDK Manager
native Ubuntu Linux 20.04 Host installed with DRIVE OS Docker Containers
native Ubuntu Linux 18.04 Host installed with DRIVE OS Docker Containers
other

Issue Description
I am trying to use pointpillar model for dnn. However, during using optimization tool im getting errors[in logs]. Does “TensorRT Optimizer Tool” handle dynamic input shape.?

Previous thread with @Morganh:
(DW_DNN_INVALID_MODEL error for trt model (isPointPillarNet | NVIDIA NGC) - #6 by atharv.sharma)

Dataset link:

point pillar documentation:
https://docs.nvidia.com/tao/tao-toolkit/text/trtexec_integration/trtexec_pointpillars.html#sample-command-for-a-pointpillars-model

Logs

Type "apropos word" to search for commands related to "word"...
Reading symbols from ./tensorRT_optimization...
(No debugging symbols found in ./tensorRT_optimization)
(gdb) run
Starting program: /usr/local/driveworks-5.14/tools/dnn/tensorRT_optimization \  --modelType=onnx \  --onnxFile=/usr/local/driveworks/samples/src/sensors/lidar/lidar_custom/model.onnx \  --out=pointpillars_trt.engine \  --half2=1 \  --workspaceSize=4096
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/usr/lib/x86_64-linux-gnu/libthread_db.so.1".
[New Thread 0x7fffbcfdb000 (LWP 1526046)]
[New Thread 0x7fffb0fde000 (LWP 1526047)]
[New Thread 0x7fff9efde000 (LWP 1526048)]
[Detaching after vfork from child process 1526049]
[Detaching after vfork from child process 1526051]
[Detaching after vfork from child process 1526053]
[Detaching after vfork from child process 1526055]
[Detaching after vfork from child process 1526057]
[Detaching after vfork from child process 1526059]
[Detaching after vfork from child process 1526061]
[Detaching after vfork from child process 1526063]
[Detaching after vfork from child process 1526065]
[Detaching after vfork from child process 1526067]
[Detaching after vfork from child process 1526069]
[Detaching after vfork from child process 1526071]
[Detaching after vfork from child process 1526073]
[Detaching after vfork from child process 1526075]
ProgramArguments: Missing argument 'outputBlobs' requested
ProgramArguments: Missing argument 'inputBlobs' requested
ProgramArguments: Missing argument 'inputDims' requested
[12-02-2025 10:15:38] WARNING: ExplicitBatch is enabled by default for ONNX models.
[12-02-2025 10:15:38] WARNING: --batchSize is ignored for explicit batch ONNX models for TRT versions after 6.2.0.3
                      Use model.batch[BATCH_SIZE].onnx (eg. `model.batch8.onnx` with ONNX batch size 8) with accompanying batch_sizes in configuration.json instead.
[12-02-2025 10:15:41] DNNGenerator: Initializing TensorRT generation on model /usr/local/driveworks/samples/src/sensors/lidar/lidar_custom/model.onnx.
[12-02-2025 10:15:41] onnx2trt_utils.cpp:372: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32.
[12-02-2025 10:15:41] DNNGenerator: Input "points": -1x204800x4
[12-02-2025 10:15:41] DNNGenerator: Input "num_points": -1
[12-02-2025 10:15:41] DNNGenerator: Output "output_boxes": -1x393216x9
[12-02-2025 10:15:41] DNNGenerator: Output "num_boxes": -1
[12-02-2025 10:15:41] Error[4]: [network.cpp::validate::3162] Error Code 4: Internal Error (Network has dynamic or shape inputs, but no optimization profile has been defined.)
[12-02-2025 10:15:41] [12-02-2025 10:15:41] Releasing Driveworks SDK Context
Error: DW_INTERNAL_ERROR: DNNGenerator: Network build and serialization failed.
[Thread 0x7fffb0fde000 (LWP 1526047) exited]
[Thread 0x7fffbcfdb000 (LWP 1526046) exited]
[Thread 0x7fffc07a7000 (LWP 1526035) exited]
[Inferior 1 (process 1526035) exited with code 01]

Dear @atharv.sharma,
I don’t find dynamic shapes related options like minShapes/maxShapes/optShapes/shapes in tensorRT_Optimization tool.
Let me check with core team and get back to you.

Dear @atharv.sharma ,
dynamic input shapes in not supported by TensorRT_Optimzation tool in DW 5.20. Please use static input.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.