Intel® Distribution of OpenVINO™ Toolkit
Community support and discussions about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all things computer vision-related on Intel® platforms.
5541 Discussions

can't use openvino gaze-estimation-adas-0002 (python)

De_Prins__Robbe
Beginner
247 Views

I'm trying to use pre-trained model gaze-estimation-adas-0002 from openvino. This link should tell me which inputs to use: https://docs.openvinotoolkit.org/2019_R1/_gaze_estimation_adas_0002_description_gaze_estimation_adas...

I detect left and right eye using haar cascades. An make blobs using:
```right_eye_image = cv.dnn.blobFromImage(right_eye, size = (60, 60), crop=False)```
```left_eye_image = cv.dnn.blobFromImage(left_eye, size = (60, 60), crop=False)```
head_pose_angles is just a list with the three headpose directions

by reading errors I know that only the first argument of model_gaze.setInput() contains a blob and that it should be of type cv::UMat. When i use ```model_gaze.setInput(cv.UMat(np.array([left_eye_image, right_eye_image, head_pose_angles])))``` I get the error ```TypeError: Required argument 'ranges' (pos 2) not found```. How do I correctly use the model?

0 Kudos
2 Replies
De_Prins__Robbe
Beginner
247 Views

Solved: https://github.com/opencv/opencv/issues/15248#issuecomment-518645343

Shubha_R_Intel
Employee
247 Views

Dear De Prins, Robbe,

Thanks for sharing your solution with the OpenVino community !

Shubha

 

Reply