eIQ Trained model used for RaspberryPi


eIQ Trained model used for RaspberryPi

700 次查看
Contributor I


I was very impressed with the eIQ tool presented in the previous webinar named "Machine Learning Development with NXP eIQ Software" thus I wanted to apply my learnings to my current hobby project based on RaspberryPi.

I created a model trained with the eIQ tool, by choosing fpn_ssd_mobilenet_v2 and ssd_mobilenet_v3 for image detection purpose. Once the model was trained successfully I saved it as a quantized tensorflow lite model but I encountered issues once I tried to run on RaspberryPi using the TFLite_detection_webcam.py from github.

"File "TFLite_detection_webcam.py", line 188, in <module>
classes = interpreter.get_tensor(output_details[1]['index'])[0] # Class index of detected objects
IndexError: list index out of range"

Reading on other websites it seems the issue is that "list index out of range" error is occurring because the exported model is using "image classification" rather than "object detection".

Could you please confirm if this is true or to propose another solution to be able to run object detection on RaspberryPi?

0 项奖励
1 回复

618 次查看
NXP Employee
NXP Employee


Please note that the EULA restricts use of eIQ Toolkit and associated software with NXP devices only
0 项奖励