Installation instruction seems not to work

[cansik]

Hi, I've just tried to install the project as described in the Installation.md and tried to run the demo command:

cd regressor

# run SHAPY regressor to estimate body shape for images in the wild
python demo.py --save-vis true --save-params true --save-mesh true --split test --datasets openpose --output-folder ../samples/shapy_fit/ --exp-cfg configs/b2a_expose_hrnet_demo.yaml --exp-opts output_folder=../data/trained_models/shapy/SHAPY_A part_key=pose datasets.pose.openpose.data_folder=../samples datasets.pose.openpose.img_folder=images  datasets.pose.openpose.keyp_folder=openpose datasets.batch_size=1 datasets.pose_shape_ratio=1.0

First of all, the demo script seems to expect a different folder structure than explained in the installation manual. It's not a problem to copy them to the right place, but with the installation script this could be automated (or adapted in the code):

# should be at data/models/smplx/
FileNotFoundError: [Errno 2] No such file or directory: '../data/expose_release/models/smplx/SMPLX_NEUTRAL.npz'

# should be at data/utility_files/smplx
FileNotFoundError: [Errno 2] No such file or directory: '../data/expose_release/utility_files/smplx/smplx_extra_joints.yaml'

And second, there are files missing, like the HRNet pretrained imagenet weights (and not mentioned in the installation):

ValueError: ../data/hrnet_v2/hrnetv2_w48_imagenet_pretrained.pth is not exist!

Would it be possible provide all the files necessary to reproduce your results? I guess it is this model: HRNet-W48-C

And libturbojpeg is needed as well (sudo apt-get install libturbojpeg).

[muelea]

Thank you for bringing this to our attention.

The smpl/smpl-x model paths are updated now and the backbone is initialised from SHAPY, because HRNet is not necessary for demo and evaluation.

At least the demo code should be running properly now.