buckeye17

I've run into the same issue. It appears that modifying line 49 of download.py as follows fixes the error (I appended `decode("utf-8")` to the method chain): `script_url = re.search(pattern, page.get_data().decode("utf-8")).group()[1:-1]`...

I eventually figured out my issues. Here's my final script to flush a dummy tensor through GaitGraph for inference: ``` import numpy as np import torch from common import get_model_resgcn...

I doubt that POEM and Pr-VIPE accidentally omitted the checkpoint files, but if this is the case, I too would like to obtain the file(s).

@B-rkh-verma after some more Googling, I discovered that the weights are actually available [here](https://sites.google.com/view/pr-vipe/model-checkpoints). It would be great if the Pr-VIPE README was updated with this link.

The Github [repo](https://github.com/JaidedAI/EasyOCR/blob/master/LICENSE) for EasyOCR has an Apache 2.0 license. Where did you find LGPL?