Dlib Face_Recognition library based Face Recognition with DeepStream

Accelerated Computing Intelligent Video Analytics DeepStream SDK
#1

Hi,

I am trying to use deepstream to detect and recognize faces using face-recognition · PyPI which uses Dlib.

The model files for face recognition using Dlib are :

dlib_face_recognition_resnet_model_v1.dat, mmod_human_face_detector.dat, shape_predictor_5_face_landmarks.dat, shape_predictor_68_face_landmarks.dat

(face_recognition_models/face_recognition_models/models at master · ageitgey/face_recognition_models · GitHub)

I am trying to convert these into caffe, uff or onnx type for using with deepstream by setting in the configuration file.

For converting .dat files to any of these types, conversion to xml seems to be the first step.

However I don’t know the exact layers in the Dlib models are. Hence I can’t parse them properly. Can you suggest any alternate method?

#3

Hey customer, it’s not a issue related to Deepstream, maybe you can create a new topic in TensorRT forum.