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.
6386 Discussions

How to feed Gaze Estimation model using OpenVINO Python API?

Sack__Alexander
Beginner
768 Views

I'd like to use the Gaze Estimation Adas 0002 pre-trained model found here:

https://docs.openvinotoolkit.org/latest/_models_intel_gaze_estimation_adas_0002_description_gaze_estimation_adas_0002.html

It takes three inputs, the left eye (1,3,60,60), right eye (1,3,60,60), and an angle vector (1,3). All of which are named parameters for the input.

However, when I try to feed these inputs to my model via:

inp_dict={'left_eye_image': inps[0], 'right_eye_image': inps[1], 'head_pose_angles': inps[2]}

input_dict={self._input_blob: inp_dict}

results = self._net.infer(input_dict)

 

OpenVINO inference barfs with it wants a float instead of a dict for the input blob. I have tried dumping the layers, iterating over input_blob (found by using next(iter(self.model.inputs)) etc.) to no avail.

How does one specify names inputs in the OpenVINO API?

0 Kudos
2 Replies
Sack__Alexander
Beginner
768 Views

Note that if I use inp_dict I get:

 

AssertionError: No input with name left_eye_image found in network

0 Kudos
Sack__Alexander
Beginner
768 Views

Nevermind, you can close this. Wrong XML was being loaded.

0 Kudos
Reply