How to use my own algorithm model to Deepstream pipeline on Xavier?

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

When I run $ ./deepstream-test3-app on Xavier , I want to use my own algorithm model.
I found the configure file : “dstest3_pgie_config.txt”.
If I want to use my own algorithm model, maybe I should change the property and replace the following files:

[property]
net-scale-factor=0.0039215697906911373
model-file=…/…/…/…/samples/models/Primary_Detector/resnet10.caffemodel
proto-file=…/…/…/…/samples/models/Primary_Detector/resnet10.prototxt
labelfile-path=…/…/…/…/samples/models/Primary_Detector/labels.txt
int8-calib-file=…/…/…/…/samples/models/Primary_Detector/cal_trt4.bin
model-engine-file=…/…/…/…/samples/models/Primary_Detector/resnet10.caffemodel_b2_int8.engine

so, here is the question:

  1. what’s detail steps should I take ,if I want to use my own algorithm model?

  2. There’re some extra processing I want to add before displaying, such as scaling, white balance and so on, is there any plug-in offered by Deepstream I can use?

  3. If no plug-in can implement those functions, could I write a plug-in, reference to
    DeepStreamSDK-Jetson-3.0_EA_beta5.0/deepstream_sdk_on_jetson/sources/gst-plugins/gst-dsexample/gstdsexample.cpp ?
    or any other suggestion?

Thanks for any reply. I’ll be appreciated.

2

Hi,
Please check the document NVIDIA_DeepStream_SDK_on_Jetson_References.
For 1, you can refer to [Application Customization]
For 2 and 3, you can refer to [Implementing a Custom Gstreamer Plugin]