How to Log Intermediate Layer Outputs in hef Model (YOLOv11)?

General
1

Hello everyone,

I am working on running inference on a YOLOv11 model using HailoRT and would like to log intermediate layer outputs. However, I noticed that hef.get_output_vstream_infos() only lists final outputs, and intermediate layers are not available by default. *I need to extract activations from specific intermediate layers.

Questions:

  1. How can I modify my inference script to log outputs from specific intermediate layers?

  2. How do I compile my YOLOv11 model to include intermediate layers as outputs?

  3. Are there best practices for logging multiple intermediate layers without significantly affecting inference speed?

2

@omria Can you please help me with this?