I have modified the pedestrian tracker demo code and in that I am detecting person and faces. For this purpose, I am using `person-detection-retail-0013` for person detection and `face-detection-retail-0004` for face detection. But the issue is that, it becomes a bit difficult sometime to check which face belongs to which person, as detection is being done by two different models.
Then I came across this model `face-person-detection-retail-002` which I believe can detect face and person. But I am unable to find how to use it. Is there any demo available which is using this model so that I can get a reference code from it.
Based on the Inference Engine Samples, the face-person-detection-retail-0002 model should work with any demo that supports SSD*-based models. For example the Hello Shape Infer SSD demo or the Object Detection for SSD Async demo. There are also some Python* samples that support SSD* based models and these are included in the toolkit under <path-to>openvino/deployment_tools/inference_engine/samples/python_samples.
Let me know if you have any additional questions.
Not sure if this is possible with the sample code provided as is, I think you may have to modify the program to be able to use the model. For your information I attempted to run the sample program (as provided) using the face-person-detection-retail-0002 model and I get an [ ERROR ] Face Detection network should have only one output.
Let me know if you have any other questions.
I am able to get the outputs for both face and person with different nodes but how to combine them ?
face_result = person_face_infer_network1.get_output(person_currReq,"detection_out_face")
pede_result = person_face_infer_network1.get_output(person_currReq,"detection_out_pedestrian")
not sure how to combine the face and person.