Can i have access to purposs build models in tao version 6?

Please provide the following information when requesting support.

• Hardware (T4/V100/Xavier/Nano/etc) a100
• Network Type (Detectnet_v2/Faster_rcnn/Yolo_v4/LPRnet/Mask_rcnn/Classification/etc) purposse build models like peoplenet
• TLT Version (Please run “tlt info --verbose” and share “docker_tag” here) tao v6 instlalaion guide
• Training spec file(If have, please share here) don’t have
• How to reproduce the issue ? (This is for errors. Please share the command line and the detailed log here.)

i wanna access to high performance models which i can use in my real-time task but i can’t see them in Performance Metrics, there was some of them like peoplenet or detectnet but they are removed from tao v6 , so now we don’t even have any purpose build model, so what should i do now? is there any high performance model still exist in tao v6?
i watch performance metrics from here

The peoplenet(actually it is trained with detectnet_v2 network) still exists in TAO 6. They are not removed. You can still use it as previously.

there wasn’t any detect_net model there in branch version 6:

i even see * Migrating to TAO 6.0 * guide which said detectnet_v2 removed entierly from tao v6

also there is another issue i can’t see any performance metric about RT-DetR
RT-DETR

Yes, you are right. You can download the docker nvcr.io/nvidia/tao/tao-toolkit:5.0.0-tf1.15.5 or nvcr.io/nvidia/tao/tao-toolkit:4.0.1-tf1.15.5 and run with docker run way. The notebook can be found in the tao_tutorials/notebooks/tao_launcher_starter_kit at tao_5.5_release · NVIDIA/tao_tutorials · GitHub instead.

I will check further internally.

then it’s not tao v6.
also you don’t answer my first question here.

May I know if you are requesting the profiling numbers of tensorrt?

my question is about fps in different nvidia graphic cards like t4 or somthing like that, i saw some of them in model zoo section but there was nothing about RT-DETR or some other models

Got it, I am still checking and seems that it is missing. Will update to you once I have.

One RT-DETR metric from a purpose-built model can be found in TrafficCamNet Transformer Lite | NVIDIA NGC.

We provide notebook for each new use case, rather than per model. Please refer to tao_tutorials/notebooks/tao_api_starter_kit/tutorials/rtdetr_detection_distillation.ipynb at main · NVIDIA/tao_tutorials · GitHub