Intel® Distribution of OpenVINO™ Toolkit
Community assistance about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all aspects of computer vision-related on Intel® platforms.
6392 Discussions

Face Detection - Accuracy checker vs Validation app

Aliyeva__Dinara
Beginner
349 Views

Hi,

 

My OpenVINO version is 2019.1.094. 

I would like to reproduce accuracy of Interactive Face Detection Demo (or, for each model in this demo).
I found that, in general, there are two ways to check model's accuracy.

Validation App - http://docs.openvinotoolkit.org/latest/_inference_engine_samples_validation_app_README.html

and Accuracy Checker tool - http://docs.openvinotoolkit.org/2019_R1/_inference_engine_tools_accuracy_checker_tool_README.html

 

My questions are:

1) Do the Accuracy checker tool and Validation app serve same purposes? 

2) Where could I be able to get all of the datasets for each model in Interactive Face Detection Demo?

Alike 'sample_dataset.tar.gz' in accuracy checker tool, but for each age/gender, emotion recognition, head-pose estimation, landmark?

3) This command:

accuracy_check -c sample/sample_config.yml -m data/test_models -s sample

doesn't work, as 

opt/intel/openvino/python/python3.5/openvino/tools/accuracy_checker/accuracy_checker/

this folder doesn't have accuracy_check built file.

It has main.py script

Then, if I run: python3 main.py -c sample/sample_config.yml -m data/test_models -s sample

it gives error that parent module is not loaded (from .config import ConfigReader)

 

Thank you

 

 

0 Kudos
1 Reply
Shubha_R_Intel
Employee
349 Views

Dear Aliyeva, Dinara,

We are phasing out the validation app so it's best not to use it anymore. It will disappear from future OpenVino releases.

Please make sure that you execute your bin/setupvars.sh[bat]

Then, you should run your accuracy_check.py from the C:\Program Files (x86)\IntelSWTools\openvino_2019.1.087\deployment_tools\tools\accuracy_checker_tool directory.

Make sure that you pip install -r C:\Program Files (x86)\IntelSWTools\openvino_2019.1.087\python\python3.6\openvino\tools\accuracy_checker\requirements.txt first.  Most of the dependencies are in the requirements.txt, though there are a few that aren't (Pillow nibabel tqdm) so you will have to pip install those individually.

To answer your question about where to get datasets for the Face Detection Demo, please navigate here:

https://github.com/opencv/training_toolbox_caffe

Thanks for using OpenVino !

Shubha

0 Kudos
Reply