search

Sindhu-Hegde / gestsync

Official code for the paper "GestSync: Determining who is speaking without a talking head" published at BMVC 2023
28 stars 1 forks source link

Not able make inference from pretrained model #1

Open sumit-chaudhary-pb15 opened 1 week ago

sumit-chaudhary-pb15 commented 1 week ago

Have completed the steps mentioned in the repo, however, not able to make inference, facing issue as below:

(p38venv_shorts) sumitc@DataPoc1:~/shorts_videos/gestsync$ python inference_activespeaker.py --checkpoint_path=checkpoints/model_rgb.pth --video_path=samples/asd_sample_1.mp4 --global_speaker=True Traceback (most recent call last): File "inference_activespeaker.py", line 20, in from protobuf_to_dict import protobuf_to_dict File "/home/sumitc@viacom18.com/shorts_videos/p38venv_shorts/lib/python3.8/site-packages/protobuf_to_dict.py", line 15, in FieldDescriptor.TYPE_INT64: long, NameError: name 'long' is not defined

Sindhu-Hegde commented 1 week ago

This is an installation issue. Can you try changing the protobuf_to_dict version and check? Also, please ensure that all the libraries installed are of the exact version as listed in the requirements.txt file.