lack of 'model_lsp.pkl' when preparing data

[hello-world-zsp]

Hi Lotayou,

Thanks for your work.

I'm following your steps to prepare my human3.6m data. I have got data washed successfully thanks to your script.

However, in PREPS.md - Create UV maps - 3. 'Run data_utils/triangulation.py'

if __name__ == '__main__':
    device=torch.device('cuda')
    data_type=torch.float32
    pose_size = 72
    beta_size = 10

    model = SMPLModel(
                device=device,
                model_path = './model_lsp.pkl',
                data_type=data_type,
                simplify=True
            )
            
    quad_obj = 'untitled.obj'
    out_obj = 'smpl_fbx_template.obj'
    fbx_obj_verts = import_verts(quad_obj)
    #model.write_obj(fbx_obj_verts, 'hybrid_fbx_verts_SMPL_face.obj')
    triangulation(quad_obj, out_obj, model.faces)

It seems that we need to obtain model_lsp.pkl first. But this pickle didn't show up in previous steps. And this pickle is also required in Step 5 and 6. I'm wondering what is it or how can I get it.

I'm quite a beginner in this field. Could you please give me more instructions?

Thank you very much.

[Lotayou]

Please refer to my previous repo for more details.

Basically, you need to download SMPL official model, and convert necessary components into a new pickle. You can use preprocess.py to achieve this.

[hello-world-zsp]

Please refer to my previous repo for more details.

Basically, you need to download SMPL official model, and convert necessary components into a new pickle. You can use preprocess.py to achieve this.

Thanks for your quick answer!

Now I'm trying to follow Step 5 to obtain my model.pkl.

But when I convert the SMPL official model , whose name is 'basicmodel_m_lbs_10_207_0_v1.0.0.pkl' , it returned a KeyError: 'cocoplus_regressor'. The keys in basicmodel_m_lbs_10_207_0_v1.0.0.pkl are: dict_keys(['J_regressor', 'weights_prior', 'f', 'v_template', 'kintree_table', 'weights', 'posedirs', 'vert_sym_idxs', 'shapedirs', 'bs_type', 'J', 'pose_training_info', 'J_regressor_prior', 'bs_style'])

So I'm not sure if I had downloaded the correct model file. Download information is:

• Address: http://smpl.is.tue.mpg.de/downloads
• Section: SMPL for Python Users
• version: 1.0.0 and the other one basicModel_f_lbs_10_207_0_v1.0.0.pkl also face the same problem.

Thank you very much.

[jinx-USTC]

Hi Lotayou,

Thanks for your work.

I'm following your steps to prepare my human3.6m data. I have got data washed successfully thanks to your script.

However, in PREPS.md - Create UV maps - 3. 'Run data_utils/triangulation.py'

if __name__ == '__main__':
    device=torch.device('cuda')
    data_type=torch.float32
    pose_size = 72
    beta_size = 10

    model = SMPLModel(
                device=device,
                model_path = './model_lsp.pkl',
                data_type=data_type,
                simplify=True
            )
            
    quad_obj = 'untitled.obj'
    out_obj = 'smpl_fbx_template.obj'
    fbx_obj_verts = import_verts(quad_obj)
    #model.write_obj(fbx_obj_verts, 'hybrid_fbx_verts_SMPL_face.obj')
    triangulation(quad_obj, out_obj, model.faces)

It seems that we need to obtain model_lsp.pkl first. But this pickle didn't show up in previous steps. And this pickle is also required in Step 5 and 6. I'm wondering what is it or how can I get it.

I'm quite a beginner in this field. Could you please give me more instructions?

Thank you very much.

Hi, can you share me the human3.6m dataset zip files? this link https://github.com/MandyMo/pytorch_HMR has expired. My email: [email protected]
Thanks~

[zhou07]

Hi Lotayou,

Thanks for your work.

I'm following your steps to prepare my human3.6m data. I have got data washed successfully thanks to your script.

However, in PREPS.md - Create UV maps - 3. 'Run data_utils/triangulation.py'

if __name__ == '__main__':
    device=torch.device('cuda')
    data_type=torch.float32
    pose_size = 72
    beta_size = 10

    model = SMPLModel(
                device=device,
                model_path = './model_lsp.pkl',
                data_type=data_type,
                simplify=True
            )
            
    quad_obj = 'untitled.obj'
    out_obj = 'smpl_fbx_template.obj'
    fbx_obj_verts = import_verts(quad_obj)
    #model.write_obj(fbx_obj_verts, 'hybrid_fbx_verts_SMPL_face.obj')
    triangulation(quad_obj, out_obj, model.faces)

It seems that we need to obtain model_lsp.pkl first. But this pickle didn't show up in previous steps. And this pickle is also required in Step 5 and 6. I'm wondering what is it or how can I get it.

I'm quite a beginner in this field. Could you please give me more instructions?

Thank you very much.

Hello, when I download human3.6m dataset from it's offical website, but I can't find the annotation which are called center, gt2d, gt3d, height, imagename, pose, shape, smpl_joint, width, so which part I need to download from offical website of human3.6m dataset?

[plmsmile]

Please refer to my previous repo for more details. Basically, you need to download SMPL official model, and convert necessary components into a new pickle. You can use preprocess.py to achieve this.

Thanks for your quick answer!

Now I'm trying to follow Step 5 to obtain my model.pkl.

But when I convert the SMPL official model , whose name is 'basicmodel_m_lbs_10_207_0_v1.0.0.pkl' , it returned a KeyError: 'cocoplus_regressor'. The keys in basicmodel_m_lbs_10_207_0_v1.0.0.pkl are: dict_keys(['J_regressor', 'weights_prior', 'f', 'v_template', 'kintree_table', 'weights', 'posedirs', 'vert_sym_idxs', 'shapedirs', 'bs_type', 'J', 'pose_training_info', 'J_regressor_prior', 'bs_style'])

So I'm not sure if I had downloaded the correct model file. Download information is:

• Address: http://smpl.is.tue.mpg.de/downloads
• Section: SMPL for Python Users
• version: 1.0.0 and the other one basicModel_f_lbs_10_207_0_v1.0.0.pkl also face the same problem.

Thank you very much.

Hey dear coder, I have the same bug as yours. KeyError: 'cocoplus_regressor'. Did you solve it? Now I am going to train a densebody model, but i meet a lot problems. I want to discuss it with you if it's convenient for you. My wechat id is plmvisionary. Thank you.

[Lotayou]

This link could be helpful, make sure you use the neutral_smpl_with_cocoplus_reg.pkl, since it contains cocoplus_regessor

[plmsmile]

This link could be helpful, make sure you use the neutral_smpl_with_cocoplus_reg.pkl, since it contains cocoplus_regessor

Thanks a lot, and I have solved this problem by your advice.

[electronicliujiang]

此链接可能会有所帮助，请确保使用neutral_smpl_with_cocoplus_reg.pkl，因为它包含cocoplus_regessor

非常感谢，我已经根据您的建议解决了这个问题。

Hello, how to solve this problem? I have not found a related .pkl file

[2294462474]

此链接可能会有所帮助，请确保使用neutral_smpl_with_cocoplus_reg.pkl，因为它包含cocoplus_regessor

非常感谢，我已经根据您的建议解决了这个问题。

Hello, how to solve this problem? I have not found a related .pkl file

https://people.eecs.berkeley.edu/~kanazawa/cachedir/hmr/models.tar.gz ,download and unzip