OS: Windows Python version: 3.8 Model: blazeface tflite model from mediapipe.
Code:
with open(model_path, 'rb') as f:
tflite_model_buf = f.read()
tflite_model = tflite.Model.GetRootAsModel(tflite_model_buf, 0)
input_tensor = "input"
input_shape = (1, 128, 128, 3)
input_dtype = "float32"
mod, params = relay.frontend.from_tflite(
tflite_model,
shape_dict = {input_tensor: input_shape},
dtype_dict = {input_tensor: input_dtype}
)
target = "llvm"
with transform.PassContext(opt_level=3):
lib=relay.build(mod, target, params=params)
I found a related error in github issues for handpose estimation, but the key was not conv2d/Bias? Can someone direct me to how I can fix this problem. It has something to do with prequantized tflite models