Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted
Beginner
11 Views

the inference_engine can't infer numpy array input data using Myriad device

Hello! I am trying to use NCS2 for inference my own model and data, my input data is a  numpy float array with the shape [1, 3, 3000, 1], instead of the unit8 format of the image data, and I am Using the script classification_sample.py with small modification, basically,it is like this:

I use:

Image = np.load(args.input)["data"]

Instead of

Image = cv2.resize(image, (w, h))

The main change is I don't use the CV module to load the input data, and it worked successfully with CPU device,but failed with MYRIAD, with the following error message:

<class 'numpy.ndarray'> (1, 3, 3000, 1)
E: [xLink] [ 4052] dispatcherEventReceive:347 dispatcherEventReceive() Read failed -4 | event 0x7fe240b68e80 XLINK_WRITE_RESP

E: [xLink] [ 4052] eventReader: 233 eventReader stopped
E: [xLink] [ 4052] XLinkReadDataWithTimeOut: 1343 Event data is invalid
E: [ncAPI] [ 4052] checkGraphMonitorResponse: 1673 XLink error, rc: X_LINK_ERROR
E: [ncAPI] [ 4052] ncGraphQueueInference: 3822 Can't get trigger response
Traceback (most recent call last):
  File "classification_sample.py", line 346, in <module>
    sys.exit(main() or 0)
  File "classification_sample.py" , line 299, in main
    res = exec_net.infer(inputs={input_blob: images}) 
  File "ie_api.pyx", line 146, in openvino.inference_engine.ie_api.ExecutableNetwork.infer
  File "ie_api.pyx", line 179, In openvino.inference_engine.ie_api.InferRequest.infer
  File "ie_api.pyx", line 183, in openvino.inference_engine.ie_api.InferRequest.infer
RuntimeError: Failed to queue inference: NC_ERROR
E: [ncAPI] [ 4195] ncFifoDestroy:3176 Failed to write to fifo before deleting it!
 

It is definitely an error related with NCS2,because I have already run successfully with CPU. Does that mean the NCS2 only works with image data? If so,that would limit its application..

Thank you in advance for your time.

0 Kudos
2 Replies
Highlighted
Beginner
11 Views

Basically,in classification_sample.py

I use:

image = np.load(args.input)["data"]

instead of :

image = cv2.imread(args.input)

And it works with CPU but fail with NCS2

0 Kudos
Highlighted
Beginner
11 Views

Now,whn I use the newest openvino_2019.3.376  ,the error message becomes:

$ python3 classification_sample_new.py -i /home/zm/generalized-phase-detection/BH_ref_answer_second/some/*.mseed -m inference_graph.xml -d MYRIAD -dt mseed
[ INFO ] Creating Inference Engine
[ INFO ] Loading network files:
    inference_graph.xml
    inference_graph.bin
[ INFO ] Preparing input blobs
[ INFO ] Batch size is 798
[ INFO ] read data time of one iteration takes: 33898.28610420227 ms :
[ INFO ] Loading model to the plugin
Traceback (most recent call last):
  File "classification_sample_new.py", line 397, in <module>
    sys.exit(main() or 0)
  File "classification_sample_new.py", line 349, in main
    exec_net = ie.load_network(network=net, device_name=args.device)
  File "ie_api.pyx", line 85, in openvino.inference_engine.ie_api.IECore.load_network
  File "ie_api.pyx", line 92, in openvino.inference_engine.ie_api.IECore.load_network
RuntimeError: AssertionFailed: _allocatedIntermData.count(topParent) > 0

As a comparison, when use CPU:

$ python3 classification_sample_new.py -i /home/zm/generalized-phase-detection/BH_ref_answer_first/*.mseed -m inference_graph.xml -d CPU -dt mseed
[ INFO ] Creating Inference Engine
[ INFO ] Loading network files:
    inference_graph.xml
    inference_graph.bin
[ INFO ] Preparing input blobs
[ INFO ] Batch size is 14704
[ INFO ] read data time of one iteration takes: 450066.1780834198 ms :
[ INFO ] Loading model to the plugin
[ INFO ] Starting inference in synchronous mode
[ INFO ] Processing output blob
orig_shape (44112000, 3)
matrix
[ INFO ] running time of one iteration takes: 34914.21580314636 ms :
[(array([   42620,    45629,    69594, ..., 44061622, 44103626, 44106614]), array([0.64882964, 0.5323102 , 0.64882964, ..., 0.64882964, 0.6478357 ,
       0.7073154 ], dtype=float32)), (array([   25397,    42612,   355297, ..., 44074285, 44092713, 44103619]), array([1.        , 1.        , 0.99999833, ..., 0.99994004, 1.        ,
       0.79882926], dtype=float32))] 2585 4918 2585 4918

Any suggestions?

0 Kudos