NVIDIA-AI-IOT / deepstream_tao_apps

Sample apps to demonstrate how to deploy models trained with TAO on DeepStream
MIT License
377 stars 97 forks source link

parsing PeopleSegNet output in deepstream with pyds NvDsObjectMeta. #63

Open Kotresh17 opened 2 years ago

Kotresh17 commented 2 years ago

Hi, I was trying to run PeopleSegNet(V2) model in deep stream, I am able to run the model fine and got the bounding box details but I am not able to access the segmentation mask output. I was trying to access with pyds NvDsObjectMeta.
obj_meta = pyds.NvDsObjectMeta.cast(l_obj.data) but I am not able to get any output from obj_meta.mask_params and I was getting error like. Screenshot (207)

Could someone please help me, how can I access the segmentation mask output.

details: Hardware Platofrm: Tesla T4 DeepStream Version: 6.0 TensorRT Version : 8.0.1 CUDA Version: 11.3 Model and config file : https://github.com/NVIDIA-AI-IOT/deepstream_tao_apps/blob/master/configs/peopleSegNet_tao/pgie_peopleSegNetv2_tao_config.txt

Thanks.