openvino could not compile blob from frozen tensorflow pb, xml or bin model - darknet

Question

openvino 2021.1 up and running
downloaded yolov3_tiny.weights and yolov3_tiny.cfg files from https://pjreddie.com/darknet/yolo/
As suggested in this link (https://colab.research.google.com/github/luxonis/depthai-ml-training/blob/master/colab-notebooks/Easy_TinyYolov3_Object_Detector_Training_on_Custom_Data.ipynb#scrollTo=2tojp0Wd-Pdw) downloaded https://github.com/mystic123/tensorflow-yolo-v3
used convert_weights_pb.py file to convert the weights and cfg file to a frozen yolov3 tiny .pb model.
python3 convert_weights_pb.py --class_names /home/user/depthai-python/my_job/coco.names --data_format NHWC --weights_file /home/user/depthai-python/my_job/yolov3-tiny.weights --tiny
used openvino mo.py file to convert yolov3_tiny .pb model to IR files .xml and .bin
python3 mo.py --input_model /home/user/depthai-python/my_job/frozen_darknet_yolov3_model.pb --tensorflow_use_custom_operations_config /home/user/depthai-python/my_job/yolo_v3_tiny.json --batch 1 --data_type FP16 --reverse_input_channel --output_dir /home/user/depthai-python/my_job
used this script as a python file to convert .xml and .bin to .blob file
blob_dir = "./my_job/"
binfile = "./my_job/frozen_darknet_yolov3_model.bin"
xmlfile = "./my_job/frozen_darknet_yolov3_model.xml"
import requests
url = "http://69.164.214.171:8083/compile" # change if running against other URL
payload = {
'compiler_params': '-ip U8 -VPU_NUMBER_OF_SHAVES 8 -VPU_NUMBER_OF_CMX_SLICES 8',
'compile_type': 'myriad'
}
files = {
'definition': open(xmlfile, 'rb'),
'weights': open(binfile, 'rb')
}
params = {
'version': '2021.1', # OpenVINO version, can be "2021.1", "2020.4", "2020.3", "2020.2", "2020.1", "2019.R3"
}
response = requests.post(url, data=payload, files=files, params=params)
print(response.headers)
print(response.content)
blobnameraw = response.headers.get('Content-Disposition')
print('blobnameraw',blobnameraw)
blobname = blobnameraw[blobnameraw.find('='):][1:]
with open(blob_dir + blobname, 'wb') as f:
f.write(response.content)
got the following error
{'Content-Type': 'application/json', 'Content-Length': '564', 'Server': 'Werkzeug/1.0.0 Python/3.6.9', 'Date': 'Fri, 09 Apr 2021 00:25:33 GMT'}
b'{"exit_code":1,"message":"Command failed with exit code 1, command: /opt/intel/openvino/deployment_tools/inference_engine/lib/intel64/myriad_compile -m /tmp/blobconverter/b9ea1f9cdb2c44bcb9bb2676ff414bf3/frozen_darknet_yolov3_model.xml -o /tmp/blobconverter/b9ea1f9cdb2c44bcb9bb2676ff414bf3/frozen_darknet_yolov3_model.blob -ip U8 -VPU_NUMBER_OF_SHAVES 8 -VPU_NUMBER_OF_CMX_SLICES 8","stderr":"stoi\n","stdout":"Inference Engine: \n\tAPI version ............ 2.1\n\tBuild .................. 2021.1.0-1237-bece22ac675-releases/2021/1\n\tDescription ....... API\n"}\n'
blobnameraw None
Traceback (most recent call last):
File "converter.py", line 29, in
blobname = blobnameraw[blobnameraw.find('='):][1:]
AttributeError: 'NoneType' object has no attribute 'find'
Alternatively i have tried the online blob converter tool from openvino http://69.164.214.171:8083/ gives the error for both .xm and .bin to .blob or from .pb to .blob
Anyone have idea.. i have tried all versions of openvino

Answer 1

We recommend you to use the myriad_compile.exe or compile_tool to convert your model to blob. Compile tool is a C++ application that enables you to compile a network for inference on a specific device and export it to a binary file. With the Compile Tool, you can compile a network using supported Inference Engine plugins on a machine that doesn't have the physical device connected and then transfer a generated file to any machine with the target inference device available.