Search code examples
tensorflowcomputer-visionyolotensorflow-lite

ValueError: Didn't find op for builtin opcode 'RESIZE_BILINEAR' version '3' Registration failed


I am using yolov4 tiny on raspberry pi4 with 2GB module.

I install the almost all packages and activate the environment.install tensor-flow and tft-lite but when i execute the detection.py code i get the error.

Value Error: Didn't find op for builtin opcode 'RESIZE_BILINEAR' version '3' Registration failed.

i am using the code of "theAIGuysCode" ...

repo link = https://github.com/theAIGuysCode/yolov4-custom-functions


Solution

  • Problem is mismatch of converter version and runtime version. Make sure you use recommended setup (tf 2.3.0):

    Conda (Recommended)

    Tensorflow CPU conda env create -f conda-cpu.yml conda activate yolov4-cpu

    Tensorflow GPU conda env create -f conda-gpu.yml conda activate yolov4-gpu

    Pip

    TensorFlow CPU pip install -r requirements.txt

    TensorFlow GPU pip install -r requirements-gpu.txt

    Check installed version (call print(tf.version) or check requirements.txt ) and install same version for your RPi.