Sorry, you must verify to complete this action. Please click the verification link in your email. You may re-send via your profile .

I am getting the error:

No module named 'openvino.runtime

after the command line: import openvino.runtime

openvino was installed on raspberry pi4 32 bit with the instructions on:

https://www.intel.com/content/www/us/en/support/articles/000057005/boards-and-kits.html

need some solution please.

Thank you for reaching out to us.

Which OpenVINO™ Toolkit version did you install on your Raspberry Pi, and did you get this error when running the OpenVINO™ Python Demo from Open Model Zoo?

For your information, I got the same error as yours when I use OpenVINO™ 2021.4.2 version and ran OpenVINO™ Python Demo from Open Model Zoo 2022.2.0 version as shown here:

Please make sure that you git clone the specific branch of Open Model Zoo that match your OpenVINO™ Toolkit package version. As mentioned, since I am using OpenVINO™ 2021.4.2, I git cloned the 2021.4.2 branch of Open Model Zoo using the command below:

git clone --depth 1 -b 2021.4.2 https://github.com/openvinotoolkit/open_model_zoo

Regards,

Megat

Thanks for the response.

I am not using a zoo model. I use my own model trained with pytorch that I convert to openvino IR using the "mo" command.
The openvino IR i get is v11.

As I mentioned above in my 1st post, I use your instruction for the raspberry pi4b:

https://www.intel.com/content/www/us/en/support/articles/000057005/boards-and-kits.html

which says i should get openvino v 2021.3 .

Please help.

Hi Farhâd,

For your information, OpenVINO™ introduced the new API 2.0 and a new OpenVINO™ Python API starting from OpenVINO™ 2022.1 releases.

The error you got is because OpenVINO™ 2021.3 uses openvino.inference_engine to Create a Core Object while OpenVINO™ API 2.0 uses openvino.runtime to Create a Core Object. You can refer to Changes to Inference Pipeline in OpenVINO API v2 for more information.

Since you have installed OpenVINO™ 2021.3, you need to use the Python inference code that implements the previous Inference Engine API. You can refer to OpenVINO™ 2021.3 Python sample codes here . You also need to convert your PyTorch model to an IRV10 model. I would suggest using the same OpenVINO™ version when converting your model.

Regards,

Megat

Thanks.

1. Can API 2.0 be installed on raspberry pi4b? If so please send me the link to the instructions.

2. I do not know how to convert your PyTorch model to an IRV10 model. Can you please refer me to a link or an instruction. I code i have used always give IRv11.

Hi Farhâd,

API 2.0 is only included in OpenVINO™ versions starting from OpenVINO™ 2022.1 release. To use API 2.0, you need to install the latest OpenVINO™ 2022 releases on your Raspberry Pi.

For your information, you can build from source OpenVINO™ 2022.1.0 on Raspberry Pi using the installation guide you mentioned. Please note that you need to git clone OpenCV 4.5.5-openvino-2022.1.0 branch and OpenVINO™ 2022.1.0 branch using the command below:

git clone --depth 1 --branch 4.5.5-openvino-2022.1.0 https://github.com/opencv/opencv.git

git clone --depth 1 --branch 2022.1.0 https://github.com/openvinotoolkit/openvino.git

Regarding your second question, I observed that you posted a similar issue in another thread. Please refer to the answer there.

Regards,

Megat

Thank you.

1.  So what I understand is that the installation guide I used for my RPi 4B is old, correct?

And that the 2 git clone commands should replace those on that guide.

2.  What about the rest of the commands in the guide? Are they up to date? For example there is a " cmake-3.14.4" command there.  Is that the correct version of cmake?

3.  How about the openCV dependencies command line:

" sudo apt install git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev libatlas-base-dev python3-scipy"

4.  Also the guide mentions Raspbian Stretch or Buster 32-bit. Are these still valid for API 2.0 installation?

5. Can you please create a new installation guide for RPi4 for API2.0? It would help avoid these issues.

Hi Farhâd,

I have built from source OpenVINO™ 2022.1 on Raspberry Pi. I'm sharing the steps below:

1. Setting up build environment.

sudo apt update && sudo apt upgrade -y

sudo apt install build-essential

2. Installing Cmake from source

Fetch CMake from the Kitware* GitHub* release page, extract it, and enter the extracted folder:

cd ~/

wget https://github.com/Kitware/CMake/releases/download/v3.18.4/cmake-3.18.4.tar.gz

tar xvzf cmake-3.18.4.tar.gz

cd ~/cmake-3.18.4

Run the bootstrap script to install additional dependencies and begin the build:

./bootstrap

make -j4

sudo make install

3. Installing OpenCV from source

Install the following packages :

sudo apt install git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev libatlas-base-dev python3-scipy

Clone the repository from OpenCV* GitHub page, prepare the build environment, and build:

cd ~/

git clone --depth 1 --branch 4.5.5-openvino-2022.1.0 https://github.com/opencv/opencv.git

cd opencv && mkdir build && cd build

cmake –DCMAKE_BUILD_TYPE=Release –DCMAKE_INSTALL_PREFIX=/usr/local ..

make -j4

sudo make install

4. Downloading source code and installing dependencies

The open-source version of Intel® OpenVINO™ toolkit is available through GitHub. The repository folder is titled openvino .

cd ~/

git clone --depth 1 --branch 2022.1.0 https://github.com/openvinotoolkit/openvino.git

Fetch the submodules from the repository:

cd ~/openvino

git submodule update --init --recursive

Run the script to install the dependencies for Intel® OpenVINO™ toolkit:

sh ./install_build_dependencies.sh

5. Building

The first step to beginning the build is telling the system where the installation of OpenCV is. Use the following command:

export OpenCV_DIR=/usr/local/lib/cmake/opencv4

To build the Python API wrapper, install all additional packages:

cd ~/openvino/samples/python/

pip3 install -r requirements.txt

The toolkit uses a CMake building system to guide and simplify the building process. To build both the inference engine and the MYRIAD plugin for Intel® Neural Compute Stick 2, use the following commands. Remove all the backslashes (\) when running the commands below. The backslashes are used to inform that these commands are not separated.

cd ~/openvino

mkdir build && cd build

cmake -DCMAKE_BUILD_TYPE=Release \

-DCMAKE_INSTALL_PREFIX=/home/pi/openvino_dist \

-DENABLE_MKL_DNN=OFF \

-DENABLE_CLDNN=OFF \

-DENABLE_GNA=OFF \

-DENABLE_SSE42=OFF \

-DTHREADING=SEQ \

-DENABLE_OPENCV=OFF \

-DNGRAPH_PYTHON_BUILD_ENABLE=ON \

-DNGRAPH_ONNX_IMPORT_ENABLE=ON \

-DENABLE_PYTHON=ON \

-DPYTHON_EXECUTABLE=$(which python3.7) \

-DPYTHON_LIBRARY=/usr/lib/arm-linux-gnueabihf/libpython3.7m.so \

-DPYTHON_INCLUDE_DIR=/usr/include/python3.7 \

-DCMAKE_CXX_FLAGS=-latomic ..

make -j4

sudo make install

For your information, I have validated these steps on Raspbian Buster 32-bit. Only Raspbian Buster 32-bit and Raspbian Stretch 32-bit are validated on OpenVINO™. We apologize for the inconvenience and will inform the relevant team to update the guide.

To uninstall your previous OpenCV build, run the command sudo make uninstall from your OpenCV 4.5.2-openvino build directory. You can refer here for more information.

Regards,

Megat

Thank you for the help.

I followed your instructions until the end.

Then I tried checking to see if cv2 can be imported and got the following error.

>>> import cv2
RuntimeError: module compiled against API version 0xe but this version of numpy is 0xd
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/pi/.local/lib/python3.7/site-packages/cv2/__init__.py", line 8, in <module>
from .cv2 import *
ImportError: numpy.core.multiarray failed to import

Seems the numpy in the requirements.txt file is not the correct version. Is that right or is it something else?

Please le me know. Thank you again.

Now a new error:

>>> import openvino
>>>
>>> import openvino.runtime as ov
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/pi/openvino_dist/python/python3.7/openvino/runtime/__init__.py", line 18, in <module>
from openvino.pyopenvino import Dimension
ImportError: libopenvino.so: cannot open shared object file: No such file or directory

cannot do import openvino.runtime

Please ignore the other errors I reported yesterday. I was able to fix them.

I need a solution to the following. I was looking for an objection detection script in the zoo samples and found the classification_sample_async.py  code.

Tried it as below and got strange numbers:

python3 ~/openvino/samples/python/classification_sample_async/classification_sample_async.py -m models/person-vehicle-bike-detection-crossroad-0078.xml -i images/walk.jpg -d MYRIAD

[ INFO ] Creating OpenVINO Runtime Core
[ INFO ] Reading the model: models/person-vehicle-bike-detection-crossroad-0078.xml
[ INFO ] Loading the model to the plugin
[ INFO ] Starting inference in asynchronous mode
[ INFO ] Image path: images/walk.jpg
[ INFO ] Top 10 results:
[ INFO ] class_id probability
[ INFO ] --------------------
[ INFO ] 673 2.0000000
[ INFO ] 589 2.0000000
[ INFO ] 785 2.0000000
[ INFO ] 533 2.0000000
[ INFO ] 540 2.0000000
[ INFO ] 652 2.0000000
[ INFO ] 659 2.0000000
[ INFO ] 547 2.0000000
[ INFO ] 722 2.0000000
[ INFO ] 568 2.0000000
[ INFO ]
[ INFO ] This sample is an API example, for any performance measurements please use the dedicated benchmark_app tool

1. I don't understand the class id's and where they come from. How can I know which id corresponds to which class?

2. The probabilities are all 2.0! Doesn't make sense.

Thank you

Hi Farhâd,

For your information, Image Classification Async Python* Sample is an inference sample for Classification, not Object Detection. To run the sample, you need to use Classification Models which are available from the Open Model Zoo Public Pre-Trained Models

For Object Detection, I would suggest you use the Object Detection Python* Demo from our Open Model Zoo demo applications. To clone the Open Model Zoo repository, you need to git clone the specific branch of Open Model Zoo that matches your OpenVINO™ Toolkit version as follows:

git clone -b 2022.1.0 https://github.com/openvinotoolkit/open_model_zoo.git

cd open_model_zoo

git submodule update --init --recursive

If you encountered the error: “ModuleNotFoundError: No Module Named ‘openvino.model_zoo.model_api’”, you need to install the Python* Model API Package as follows:

pip install <omz_dir>/demos/common/python

On another note, here is the list of Supported Models for the Object Detection Python Demo.

Regards,

Megat

Installed as above with Model API.

Ran the Object Detection Python Demo above but get errors. Could be because my model is built with YOLOv5?

Here is what I got:

python3 /home/pi/open_model_zoo/demos/object_detection_demo/python/object_detection_demo.py -m models/<xml model> -i images/<image> -at yolo -d MYRIAD

[ INFO ] OpenVINO Runtime
[ INFO ] build: custom_HEAD_cdb9bec7210f8c24fde3e416c7ada820faaaa23e
[ INFO ] Reading model models/best_10-28-22_epoch=100.xml
[ WARNING ] The parameter "input_size" not found in YOLO wrapper, will be omitted
[ INFO ] Input layer: images, shape: [1, 3, 512, 512], precision: f32, layout: NCHW
[ INFO ] Output layer: output0, shape: [1, 16128, 12], precision: f32, layout:
[ INFO ] The model models/best_10-28-22_epoch=100.xml is loaded to MYRIAD
[ INFO ] Device: MYRIAD
[ INFO ] Number of streams: MYRIAD_THROUGHPUT_STREAMS_AUTO
[ INFO ] Number of model infer requests: 5
Traceback (most recent call last):
File "/home/pi/open_model_zoo/demos/object_detection_demo/python/object_detection_demo.py", line 298, in <module>
sys.exit(main() or 0)
File "/home/pi/open_model_zoo/demos/object_detection_demo/python/object_detection_demo.py", line 260, in main
results = detector_pipeline.get_result(next_frame_id_to_show)
File "/home/pi/.local/lib/python3.7/site-packages/openvino/model_zoo/model_api/pipelines/async_pipeline.py", line 124, in get_result
result = self.model.postprocess(raw_result, preprocess_meta), {**meta, **preprocess_meta}
File "/home/pi/.local/lib/python3.7/site-packages/openvino/model_zoo/model_api/models/yolo.py", line 123, in postprocess
detections = self._parse_outputs(outputs, meta)
File "/home/pi/.local/lib/python3.7/site-packages/openvino/model_zoo/model_api/models/yolo.py", line 225, in _parse_outputs
detections += self._parse_yolo_region(out_blob, meta['resized_shape'], layer_params[1])
File "/home/pi/.local/lib/python3.7/site-packages/openvino/model_zoo/model_api/models/yolo.py", line 131, in _parse_yolo_region
predictions = permute_to_N_HWA_K(predictions, params.bbox_size, params.output_layout)
File "/home/pi/.local/lib/python3.7/site-packages/openvino/model_zoo/model_api/models/yolo.py", line 41, in permute_to_N_HWA_K
assert tensor.ndim == 4, tensor.shape
AssertionError: (1, 16128, 12)

Hi Farhâd,

Thanks for your information.

For your information, we regret to inform you that YOLOv5 is not a Supported Models in Object Detection Python Demo .

The supported models for Object Detection Python Demo are shown as follow:

· architecture_type = centernet

ctdet_coco_dlav0_512

· architecture_type = ctpn

ctpn

· architecture_type = detr

detr-resnet50

· architecture_type = faceboxes

faceboxes-pytorch

· architecture_type = nanodet

nanodet-m-1.5x-416

· architecture_type = nanodet-plus

nanodet-plus-m-1.5x-416

· architecture_type = retinaface-pytorch

retinaface-resnet50-pytorch

· architecture_type = ssd

efficientdet-d0-tf

efficientdet-d1-tf

face-detection-0200

face-detection-0202

face-detection-0204

face-detection-0205

face-detection-0206

face-detection-adas-0001

face-detection-retail-0004

face-detection-retail-0005

face-detection-retail-0044

faster-rcnn-resnet101-coco-sparse-60-0001

faster_rcnn_inception_resnet_v2_atrous_coco

faster_rcnn_resnet50_coco

pedestrian-and-vehicle-detector-adas-0001

pedestrian-detection-adas-0002

pelee-coco

person-detection-0106

person-detection-0200

person-detection-0201

person-detection-0202

person-detection-0203

person-detection-0301

person-detection-0302

person-detection-0303

person-detection-retail-0013

person-vehicle-bike-detection-2000

person-vehicle-bike-detection-2001

person-vehicle-bike-detection-2002

person-vehicle-bike-detection-2003

person-vehicle-bike-detection-2004

product-detection-0001

retinanet-tf

rfcn-resnet101-coco-tf

ssd300

ssd512

ssd_mobilenet_v1_coco

ssd_mobilenet_v1_fpn_coco

ssd-resnet34-1200-onnx

ssdlite_mobilenet_v2

vehicle-detection-0200

vehicle-detection-0201

vehicle-detection-0202

vehicle-detection-adas-0002

vehicle-license-plate-detection-barrier-0106

vehicle-license-plate-detection-barrier-0123

· architecture_type = ultra_lightweight_face_detection

ultra-lightweight-face-detection-rfb-320

ultra-lightweight-face-detection-slim-320

· architecture_type = yolo

mobilefacedet-v1-mxnet

mobilenet-yolo-v4-syg

person-vehicle-bike-detection-crossroad-yolov3-1020

yolo-v1-tiny-tf

yolo-v2-ava-0001

yolo-v2-ava-sparse-35-0001

yolo-v2-ava-sparse-70-0001

yolo-v2-tf

yolo-v2-tiny-ava-0001

yolo-v2-tiny-ava-sparse-30-0001

yolo-v2-tiny-ava-sparse-60-0001

yolo-v2-tiny-tf

yolo-v2-tiny-vehicle-detection-0001

yolo-v3-tf

yolo-v3-tiny-tf

· architecture_type = yolov3-onnx

yolo-v3-onnx

yolo-v3-tiny-onnx

· architecture_type = yolov4

yolo-v4-tf

yolo-v4-tiny-tf

· architecture_type = yolof

yolof

· architecture_type = yolox

yolox-tiny

Hope it helps.

Regards,

Wan

Thanks for your patience.

The error you encountered: “ AssertionError: (1, 16128, 12) ” was due to your model’s output shape being (1, 16128, 12) while tensor.ndim require 4, tensor.shape .

To solve this issue, you must convert your ONNX model into Intermediate Representation with the following command:

mo --input_model <ONNX_model> --output <output_names> --data_type FP16 --scale_values=<input_names>[255] --input_shape= <input_shape> --input=<input_names>

Thank you.

1. Could you please describe the options and give some examples so i could understand better?

2. Originally I converted my pytorch model to onnx and IR.  Could I have miss something in that conversion that caused the dimensionality problem. Like missing an option?

How do you install "mo" on raspberry pi4B?

I have openvino 2022.1 installed. Python version 3.7.3

I tried the command :

>> pip install openvino-dev[onnx]

But got the error:

ERROR: Cannot install openvino-dev[onnx]==2021.3.0, openvino-dev[onnx]==2021.4.0, openvino-dev[onnx]==2021.4.1, openvino-dev[onnx]==2021.4.2, openvino-dev[onnx]==2022.1.0 and openvino-dev[onnx]==2022.2.0 because these package versions have conflicting dependencies.

The conflict is caused by:
openvino-dev[onnx] 2022.2.0 depends on rawpy~=0.17.1; python_version >= "3.7"
openvino-dev[onnx] 2022.1.0 depends on openvino==2022.1.0
openvino-dev[onnx] 2021.4.2 depends on openvino==2021.4.2
openvino-dev[onnx] 2021.4.1 depends on rawpy>=0.16.0
openvino-dev[onnx] 2021.4.0 depends on rawpy>=0.16.0
openvino-dev[onnx] 2021.3.0 depends on fast-ctc-decode>=0.2

Community support is provided during standard business hours (Monday to Friday 7AM - 5PM PST). Other contact methods are available here .

Intel does not verify all solutions, including but not limited to any file transfers that may appear in this community. Accordingly, Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.

For more complete information about compiler optimizations, see our Optimization Notice .