Ds-example custom plugin

**• Hardware Platform- Jetson Xavier
**• DeepStream Version- 6.1
**• Jetpack Version- 5.0.2

I have an anomaly detection model that produces an image as output. I aim to measure the dissimilarity between the input image (current frame) and the model’s output. Currently, I am using the following pipeline: v4l2src → nvpreprocess (ROI) → nvinfer → dsexample. Within the dsexample plugin, I am modifying the blur_objects function. My approach involves post-processing the output image first and then computing the difference between the input image and the post-processed output. I am curious if this method is suitable or if there are alternative approaches to achieve the same goal. Also, can I use queue in my pipeline to send the current frame to nvinfer as well as dsexample plugin?

Your implementation is OK.

What do you mean?

How can I retrieve the current original frame as well as the output image frame from the nvinfer plugin within my custom ds-example plugin?

hey, any solution for accessing the current original frame as well as output image frame from the nvinfer plugin within ds-example plugin?

The original frames can be available in NvBufSurface.

There is a simple sample for how to map the Jetson platform NvBufSurface buffer. deepstream_tao_apps/apps/tao_others/deepstream-heartrate-app/heartrateinfer_impl/heartrateinfer.cpp at master · NVIDIA-AI-IOT/deepstream_tao_apps · GitHub, deepstream_tao_apps/apps/tao_others/deepstream-heartrate-app/heartrateinfer_impl/heartrateinfer.cpp at master · NVIDIA-AI-IOT/deepstream_tao_apps · GitHub and deepstream_tao_apps/apps/tao_others/deepstream-heartrate-app/heartrateinfer_impl/heartrateinfer.cpp at master · NVIDIA-AI-IOT/deepstream_tao_apps · GitHub

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.