Is it possible at all?
My first problem was index out of range
which can be solved by simplifying the only model. Next, you are faced with unsupported layers, which can be solved using recommended end node names that are different from the expected ones. Okay… Next, it was yolov8n/conv41 didn’t exist, and developers said that models with a small number of classes could erase some of and adding the line quantization_param([conv42, conv53, conv63], force_range_out=[0.0, 1.0])
can solve this problem. Still, I see that conv41, not yolov8n/conv41, solved my problem, and as the fact after I changed all files as it was in the model_zoo repo, I had ValueError: Exception encountered when calling layer "yolov8_nms_postprocess" (type HailoPostprocess)
. Now, I have an Error in the last layers of the model, expected conv but found LayerType.activation layer
. What do you recommend me to do? Thank you!
Hi @ighgul,
I would like to take a look at the full picture, do you think that I can use the same dataset as you’ve used, or at least similar?
Ideally this should have been much smoother
Thank you a lot. I fix these problems converted by terminal using hailo_model_zoo main.py