Not able to run onnx model in deepstream

[clara.goncalves]

**• Hardware Platform Jetson nano **
• DeepStream Version 6.0
• JetPack Version 4.6.2
• TensorRT Version 8.2.1.8
• not able to run onnx model in deepstream

How can I can addapt the config files from deepstream-app to run with the current output from nvidia tao toolkit:

• resnet18_detector.onnx
• calibration.bin
• resnet18_detector.trt.int8
  When I try to run with the onnx-model the output vedeo file doesn’t detect any objects. But while trainning the model it gave an 66% accuracy. The model was also trained on top of a previous trained model in .tlt .

[junshengy]

How did you run the model ？
Can you share you model and deepstream configuration file ?

[clara.goncalves]

I am only able to provide the configuration file, but I’m experiencing difficulty in successfully runt he deepstram app with any ONNX model.
config.txt (448 Bytes)

[junshengy]

If you add the following item to you configuration file, can the program run normally?

int8-calib-file="your int8 calibration file"

you also can add this item to save the converted model.

model-engine-file="converted model"

Was some error happened when load the model ?

[clara.goncalves]

Upon integrating the int8-calib file and model-engine file, I encountered the same issue where running the deepstream-app successfully did not result in object detection.

[junshengy]

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