Object Detection Model Serving Not Working in OpenVINO about openvino HOT 9 CLOSED

@andrei-kochin thank you so much.

from openvino.

thanks for letting us know
Could you share a model?
Could you try 2024.2 version to compare behavior?
Are you trying to import pytorch model (as report states), or converted to ONNX (by the code)?

cc/ @AKochin

from openvino.

@mlukasze , Could you kindly review the issue description thoroughly?

Please find the model attached: yolov10n_openvino_model.zip. I am attempting to infer the OpenVINO model but am encountering an issue with version 2024.2.

Thank you for your assistance.

from openvino.

thanks for the explanation, our engineers will look at your problem in first possible moment.
we will back to you

from openvino.

@dsnsabari according to the code snippet provided and to the error you see you are trying to feed the read_model call the directory instaead of the IR *.xml file which is not expected. Was this code really working fine with 22.3?

from openvino.

@andrei-kochin , I am not using this code to read the model in the 22.3 version. Could you share the sample script for serving the open vino model in the latest package?

from openvino.

@dsnsabari I'm not the expert in serving but I definitely can say that your code for the read_model should look like this to work:

converted_model_path = "./yolov8n_openvino_model/yolov8n.xml"

# Read the network and corresponding weights from a file.
model = core.read_model(model=converted_model_path)

Or you even can skip read_model and feed it directly to the compile_model():

converted_model_path = "./yolov8n_openvino_model/yolov8n.xml"

# Compile the model for CPU (you can choose manually CPU, GPU etc.)
# or let the engine choose the best available device (AUTO).
compiled_model = core.compile_model(model=converted_model_path, device_name=device.value)

from openvino.

@andrei-kochin , Thank you so much. I want to set the config parameters below while initializing the model. "AttributeError: 'Core' object has no attribute 'set_config'"

Old openvino serving code :

from inference_engine import IECore

# Configure device
ie = IECore()
config = {"CPU_THREADS_NUM": "2", "CPU_BIND_THREAD": "YES"}
ie.set_config(config, "CPU")  # Replace "CPU" with your target device

# Load network and perform inference
# ... (Your inference code here)

Please help me config the parameters for the new openvino:

core = ov.Core()
#config = {"CPU_THREADS_NUM": "2", "CPU_BIND_THREAD": "YES"}
core.set_config({"CPU_THREADS_NUM": "2", "CPU_BIND_THREAD": "YES"}, "CPU")

from openvino.

@dsnsabari new call is the set_property. Please see the doc reference.

Some examples can be found here.

from openvino.