search

yiranran / Audio-driven-TalkingFace-HeadPose

Code for "Audio-driven Talking Face Video Generation with Learning-based Personalized Head Pose" (Arxiv 2020) and "Predicting Personalized Head Movement From Short Video and Speech Signal" (TMM 2022)
https://ieeexplore.ieee.org/document/9894719
721 stars 146 forks source link

RuntimeError: Unable to open ../Deep3DFaceReconstruction/shape_predictor_68_face_landmarks.dat #41

Open ghost opened 3 years ago

ghost commented 3 years ago

RuntimeError: Unable to open ../Deep3DFaceReconstruction/shape_predictor_68_face_landmarks.dat where is the shape_predictor_68_face_landmarks.dat? it is not in Deep3DFaceReconstruction.

jordanfranklin1 commented 3 years ago

Hey, not sure if you found out already but I had the same issue.

shape_predictor_68_face_landmarks.dat is located in the Deep3DFaceReconstruction folder BUT in the folder of the models that you would have to download from their Baidu or Google Drive link. There is also a Deep3DFaceReconstruction folder in this repo BUT shape_predictor_68_face_landmarks.dat will not be there.

I guess the best way is to download the repo AND the models folder from either the Baidu or Google Drive link, copy shape_predictor_68_face_landmarks.dat from the Deep3DFaceReconstruction folder in the models folder, and paste it into the Deep3DFaceReconstruction folder in this repo to run the extract_frame1.py script successfully.