Creating engine for yolov4_tiny on jetson xavier nx

[sarvatmak]

tools used for generating qat enabled yolov4_tiny module
1 Hardware GeForce RTX 4070 Ti
2 Network Type (yolov4_tiny )
3 TLT Version (format_version: 2.0, toolkit_version: 4.0.1)

generated files are

• .etlt file
• cal.json file
  and they are trained by enabling QAT.

link to my previous question for overview is

---

Possibility of QAT training for Jetson devices for yolov4_tiny model with pruned etlt model

• Hardware GeForce RTX 4070 Ti • Network Type (Detectnet_v2) • TLT Version (format_version: 2.0, toolkit_version: 4.0.1) This is link to my previous question for more reference Very low precision while Training detectnet_v2 model using custom data in TAO Now I have got considerable accuracy of about 0.79 mAP and my aim is to deploy this model on jetson xavier NX device in deepstream. I have generated following files in export folder cal.bin -------- calibration cache file cal.tensorfi…

---

Now i want to deploy the saved model on jetson

• Hardware Platform : Jetson xavier nx
• DeepStream Version : 6.2
• JetPack Version : 5.1

Now since it is QAT trained model i can only deploy it in int8 mode(correct?).
I am not able to get an idea how to do it.
please explain the procedure i am not able to build tensrrt oss with docs provided.
most probably because tensorrt i have is 8.5.2.2 and it is downloading older one,now i am trying release versions (8.5.2) file but still it is not clear to me.
any more help will be appreciated.
Also it says CMAKE_CUDA_COMPILER could be found.

[yuweiw]

What do you mean below?

@sarvatmak

i am not able to build tensrrt oss with docs provided.

[sarvatmak]

sorry i was trying to generate tensorrt oss for tensorrt 8.5.2 using old tensorrt version i successfully generated the file by making necessary changes (i was using tensorrt 7 repo).
thanks anyways.