Can you upload the data on google drive or huggingface?

[bareblackfoot]

I cannot download files from baidu :(

[Ryan-ZL-Lin]

Hi Is there anyone could provide the Pre-trained models to Google Drive, please? I cannot download the file from Baidu either.

[dennisushi]

Hi, @minghanqin , could you look into this?

[minghanqin]

Hi,@bareblackfoot @dennisushi @Ryan-ZL-Lin Thank you for your attention! The google drive link is updated now.

[Ryan-ZL-Lin]

@minghanqin Thanks a lot, I downloaded the pretrained model from Google Drive successfully. But I got a new issue when using these pretrained models and not sure whether I'm using these models in the right way.

I followed the Quickstart to get started by using sofa_1 in the downloaded files. After running python render.py -m output/sofa_1 --include_feature, I encountered an error saying AssertionError: Could not recognize scene type!, here is the full error log:

(langsplat) ryan_lin@RyanLegionPro7i:/mnt/c/Users/lzlal/LangSplat$ python render.py -m output/sofa/sofa_1 --include_feature
Looking for config file in output/sofa/sofa_1/cfg_args
Config file found: output/sofa/sofa_1/cfg_args
Rendering output/sofa/sofa_1
/home/qmh/code/gaussian-splatting/download/new_datasets/sofa [18/03 16:09:36]
Traceback (most recent call last):
  File "render.py", line 88, in <module>
    render_sets(model.extract(args), args.iteration, pipeline.extract(args), args.skip_train, args.skip_test, args)
  File "render.py", line 57, in render_sets
    scene = Scene(dataset, gaussians, shuffle=False)
  File "/mnt/c/Users/lzlal/LangSplat/scene/__init__.py", line 50, in __init__
    assert False, "Could not recognize scene type!"
AssertionError: Could not recognize scene type!

After checking the code, it seems that the scene initialization is expecting either a sparse folder or a transform_train.json file, but I cannot find either one of them in the downloaded files (sofa_1, sofa_2, and sofa_3).

[wrencanfly]

@minghanqin Thanks a lot, I downloaded the pretrained model from Google Drive successfully. But I got a new issue when using these pretrained models and not sure whether I'm using these models in the right way.

I followed the Quickstart to get started by using sofa_1 in the downloaded files. After running python render.py -m output/sofa_1 --include_feature, I encountered an error saying AssertionError: Could not recognize scene type!, here is the full error log:

(langsplat) ryan_lin@RyanLegionPro7i:/mnt/c/Users/lzlal/LangSplat$ python render.py -m output/sofa/sofa_1 --include_feature
Looking for config file in output/sofa/sofa_1/cfg_args
Config file found: output/sofa/sofa_1/cfg_args
Rendering output/sofa/sofa_1
/home/qmh/code/gaussian-splatting/download/new_datasets/sofa [18/03 16:09:36]
Traceback (most recent call last):
  File "render.py", line 88, in <module>
    render_sets(model.extract(args), args.iteration, pipeline.extract(args), args.skip_train, args.skip_test, args)
  File "render.py", line 57, in render_sets
    scene = Scene(dataset, gaussians, shuffle=False)
  File "/mnt/c/Users/lzlal/LangSplat/scene/__init__.py", line 50, in __init__
    assert False, "Could not recognize scene type!"
AssertionError: Could not recognize scene type!

After checking the code, it seems that the scene initialization is expecting either a sparse folder or a transform_train.json file, but I cannot find either one of them in the downloaded files (sofa_1, sofa_2, and sofa_3).

hi, did you fix the problem?

[minghanqin]

Hi@Ryan-ZL-Lin @wrencanfly ,

They are in the Preprocessed Dataset. Just download them from the link in the README.

[daniekpo]

@minghanqin Thanks a lot, I downloaded the pretrained model from Google Drive successfully. But I got a new issue when using these pretrained models and not sure whether I'm using these models in the right way. I followed the Quickstart to get started by using sofa_1 in the downloaded files. After running python render.py -m output/sofa_1 --include_feature, I encountered an error saying AssertionError: Could not recognize scene type!, here is the full error log:

(langsplat) ryan_lin@RyanLegionPro7i:/mnt/c/Users/lzlal/LangSplat$ python render.py -m output/sofa/sofa_1 --include_feature
Looking for config file in output/sofa/sofa_1/cfg_args
Config file found: output/sofa/sofa_1/cfg_args
Rendering output/sofa/sofa_1
/home/qmh/code/gaussian-splatting/download/new_datasets/sofa [18/03 16:09:36]
Traceback (most recent call last):
  File "render.py", line 88, in <module>
    render_sets(model.extract(args), args.iteration, pipeline.extract(args), args.skip_train, args.skip_test, args)
  File "render.py", line 57, in render_sets
    scene = Scene(dataset, gaussians, shuffle=False)
  File "/mnt/c/Users/lzlal/LangSplat/scene/__init__.py", line 50, in __init__
    assert False, "Could not recognize scene type!"
AssertionError: Could not recognize scene type!

After checking the code, it seems that the scene initialization is expecting either a sparse folder or a transform_train.json file, but I cannot find either one of them in the downloaded files (sofa_1, sofa_2, and sofa_3).

hi, did you fix the problem?

@minghanqin Thanks a lot, I downloaded the pretrained model from Google Drive successfully. But I got a new issue when using these pretrained models and not sure whether I'm using these models in the right way. I followed the Quickstart to get started by using sofa_1 in the downloaded files. After running python render.py -m output/sofa_1 --include_feature, I encountered an error saying AssertionError: Could not recognize scene type!, here is the full error log:

(langsplat) ryan_lin@RyanLegionPro7i:/mnt/c/Users/lzlal/LangSplat$ python render.py -m output/sofa/sofa_1 --include_feature
Looking for config file in output/sofa/sofa_1/cfg_args
Config file found: output/sofa/sofa_1/cfg_args
Rendering output/sofa/sofa_1
/home/qmh/code/gaussian-splatting/download/new_datasets/sofa [18/03 16:09:36]
Traceback (most recent call last):
  File "render.py", line 88, in <module>
    render_sets(model.extract(args), args.iteration, pipeline.extract(args), args.skip_train, args.skip_test, args)
  File "render.py", line 57, in render_sets
    scene = Scene(dataset, gaussians, shuffle=False)
  File "/mnt/c/Users/lzlal/LangSplat/scene/__init__.py", line 50, in __init__
    assert False, "Could not recognize scene type!"
AssertionError: Could not recognize scene type!

After checking the code, it seems that the scene initialization is expecting either a sparse folder or a transform_train.json file, but I cannot find either one of them in the downloaded files (sofa_1, sofa_2, and sofa_3).

hi, did you fix the problem?

@minghanqin Thanks a lot, I downloaded the pretrained model from Google Drive successfully. But I got a new issue when using these pretrained models and not sure whether I'm using these models in the right way. I followed the Quickstart to get started by using sofa_1 in the downloaded files. After running python render.py -m output/sofa_1 --include_feature, I encountered an error saying AssertionError: Could not recognize scene type!, here is the full error log:

(langsplat) ryan_lin@RyanLegionPro7i:/mnt/c/Users/lzlal/LangSplat$ python render.py -m output/sofa/sofa_1 --include_feature
Looking for config file in output/sofa/sofa_1/cfg_args
Config file found: output/sofa/sofa_1/cfg_args
Rendering output/sofa/sofa_1
/home/qmh/code/gaussian-splatting/download/new_datasets/sofa [18/03 16:09:36]
Traceback (most recent call last):
  File "render.py", line 88, in <module>
    render_sets(model.extract(args), args.iteration, pipeline.extract(args), args.skip_train, args.skip_test, args)
  File "render.py", line 57, in render_sets
    scene = Scene(dataset, gaussians, shuffle=False)
  File "/mnt/c/Users/lzlal/LangSplat/scene/__init__.py", line 50, in __init__
    assert False, "Could not recognize scene type!"
AssertionError: Could not recognize scene type!

After checking the code, it seems that the scene initialization is expecting either a sparse folder or a transform_train.json file, but I cannot find either one of them in the downloaded files (sofa_1, sofa_2, and sofa_3).

hi, did you fix the problem?

The problem is in the cfgargs file in sofa*. The source_path is hardcoded in the file, you have to modify it to point to the 3D-OVS sofa dataset (or whichever dataset you're rendering the output for).