Deploy DINO on Jetpack 5.1.2

Autonomous Machines Jetson & Embedded Systems Jetson Orin NX
1

Hello,
I’ve exported my trained DINO detection model to onnx. I’ve followed the instructions in “Deploying to DeepStream for DINO - NVIDIA Docs” but I get the same error when building the engine file as this user encountered in this post:
“TAO 5.0 Classification (PyTorch) deploy error - #21 by 77438441”

SSA validation FAIL: tensor __mye53 has multiple definitions
peephole.cpp:657: DCHECK(g_->ssa_validation()) failed.

I’m on Jetpack 5.1.2 which has TensorRT version 8.5.2. Their issue was fixed when they upgraded to version 8.5.3. This is not possible for me since its not possible to upgrade TensortRT versions on Jetson. I also can’t upgrade to Jetpack 6.0 DP since I need Ubuntu 20.04.

Has anyone successfully deployed DINO on Jetson with a Jetpack version < 6.0?
Am I out of luck or do the planned Jetpack 5.1.3/4 versions include newer TensorRT versions?

TensorRT Version: 8.5.2
GPU Type: NVIDIA Jetson Orin NX (16GB)
Nvidia Driver Version:
CUDA Version: 11.4.315
CUDNN Version: 8.6.0.166
Operating System + Version: Ubuntu 20.04

1 Like
2

I am experiencing the same issue. Could you please assist in resolving it?

3

Hi @cznle @dasr ,
Moving this to Jetson Forum for better asssiatnce

Thanks

4

There is no update from you for a period, assuming this is not an issue any more.
Hence we are closing this topic. If need further support, please open a new one.
Thanks

Sorry for the late response.
Is this still an issue to support? Any result can be shared?