supplyiop.blogg.se

In older threads I read about a permute layer which could lead to this issues, but I can see a layer like that in my. The model is read successfully but unfortunately the program crashes with this message: Traceback (most recent call last):Ĭv2.error: OpenCV(4.1.0-openvino) /home/jenkins/workspace/OpenCV/OpenVINO/build/opencv/modules/dnn/src/op_inf_engine.cpp:747: error: (-215:Assertion failed) Failed to initialize Inference Engine backend: AssertionFailed: perm.size() = 4 in function 'initPlugin' # Prepare input blob and perform an inference.īlob = cv.dnn.blobFromImage(frame, size=(224, 224), ddepth=cv.CV_8U) Net.setPreferableTarget(cv.dnn.DNN_TARGET_MYRIAD)įrame = cv.imread('/home/pi/projects/02_ncs_test/orange_89.jpg') bin to the RPI and run the following Python code: import numpy as np Inference is run on a RaspberryPi3B+, so I copy the. My custom model consists of a Keras and Tensorflow part, whereas the last layer is a Reshape layer.

I'm using OpenVino v2019.1.144 and successfully converted my trained Tensorflow graph to IR using: python3 mo_tf.py -input_model /modelpb/encoder.pb -input_shape -data_type FP16 Unfortunately, I'm currently struggling with an issue I couldn't resolve so far and couldn't find in this form. I just started playing around with the NCS2 and could fix a lot of potential pitfalls thanks to this excellent forum.