Deploying a r3d_18 model with raspberry pi 5 + hailo 8

i’m learning and testing the hailo 8 on raspberry pi 5, and i find the r3d_18 model for video recognition has been supported by hailo 8, but the community is lack of an example of deploying this model in GStreamer. Could someone give me an example code of how to use this model in GStreamer?

Hey @RAF_WANG,

Welcome to the Hailo Community!

To get started, I recommend using the simple detection example as your foundation. You can customize it by:

  • Replacing the HEF file with your own model
  • Modifying the post-processing logic
  • Implementing your desired callback functions for handling results

Here’s the example code that will run your HEF: hailo-apps-infra/hailo_apps/hailo_app_python/apps/detection_simple/detection_pipeline_simple.py at main · hailo-ai/hailo-apps-infra · GitHub

Hope this helps!

Thank you for your reply.

I have already read all official guides and had the experience on the video recognition task using Jetson Orin + Deepstream. I’ll dive deep into Rapsberry pi + Hailo, once I work out a working example, I’ll share it with Hailo Community.:handshake:

Hi @RAF_WANG

Here is an example of running r3d_18 model using DeGirum PySDK: hailo_examples/examples/028_video_classification.ipynb at main · DeGirum/hailo_examples. Hope this helps.

Thank you so much, I’ll study this material.