From your log,
Can you try to modify the output blob name to deepstream_tao_apps/configs/nvinfer/bodypose2d_tao/bodypose2d_pgie_config.yml at release/tao5.3_ds7.0ga · NVIDIA-AI-IOT/deepstream_tao_apps · GitHub?
I have already done that and still same issue.
HI there you havent replied yet but anyways here is another question from my side.
Since iam working with COCO 2017 keypoints and the model is initally trained and pipeline is setup for 18 keypoints.I have fine-tuned the dataset but is there anything else that is causing the issues?
Have a look at below file being passed inside yaml file for inferencing.
model-config:
config-file-path: ../../../configs/nvinfer/bodypose2d_tao/sample_bodypose2d_model_config.yml
Lastly please also let me know if infer-dims parameter inside config file also effects the piepline and causes this invalid paf dimension issue?
That’s a question back to Deepstream. Could you try to modify deepstream_tao_apps/configs/nvinfer/bodypose2d_tao/sample_bodypose2d_model_config.yml at release/tao5.3_ds7.0ga · NVIDIA-AI-IOT/deepstream_tao_apps · GitHub to reflect your changing?
Hi there i updated this file and made key pairs to 32 previously by default they were 38.
That issues has been resolved now only issue is with Heatmap dimensions.See log values for model layers below:
and the error log is as below:
PLease let me know what to change for heatmap issue in detail.
initially the model that was working by default has 19 value for heatmap and for paf maping it has 38 but since my model has 18 and 32 so now let me know what to set for heatmap.
This proves that there should be something changing in deepstream side. So, it is back to a question for deepstream config/setting/etc. Could you raise question to deepstream forum for further checking? Thanks.
There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks