- Hardware: Jetson AGX Orin
- Deesptream7.0
- Jetpack 6.0 (6.0+b106 and 6.0+b87 both are installed) L4T 36.3.0
- TensorRT 8.6.2
- NVRM version: NVIDIA UNIX Open Kernel Module for aarch64 540.3.0
- Issue : I want to know how can i run a pre-trained facial landmark using the provided TAO sample application for facial landmarks, i have referred to the following BYOM documentation: Bring Your Own Model (BYOM) - NVIDIA Docs but i am not able to find out if my model is compatible or not.
- here is the model i am using: It is named
2d106detinsightface/alignment/coordinate_reg at master · deepinsight/insightface · GitHub
Since Deepstream can recognize the model 2d106det-0000.params in 2d106det, can the model be converted to ONNX format? if yes, you can modify the DeepStream sample deepstream-faciallandmark-app to customize.
yes as my model is already in onnx format so i just need to modify the sample app?
yes. the first model is used to detect faces. you can replace the second model with you own model. especially you make sure the preprocess parameters are correct. please refer to the document for the parameter explanation. about postprocessing, you can process the inference results and add meta in sgie_pad_buffer_probe by referring to the existing logics.
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
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.