gap-lab-cuhk-sz / rfdnet Goto Github PK

Hello, thanks for the amazing work!

I'm trying to reproduce the results with the pre-trained model, but I got quite different per category AP scores from the paper:

|           | display | bathtub | trashbin | sofa  | chair | table | cabinet | bookshelf | mAP   |
| --------- | ------- | ------- | -------- | ----- | ----- | ----- | ------- | --------- | ----- |
| paper     | 26.67   | 27.57   | 23.34    | 15.71 | 12.23 | 1.92  | 14.48   | 13.39     | 16.90 |
| reproduce | 23.13   | 15.89   | 18.00    | 41.61 | 10.13 | 0.95  | 26.35   | 9.10      | 18.14 |

Besides, there seems to be a lot of false positives at conf_thresh = 0.05:

----------iou_thresh: 0.500000----------
[eval mesh] table
[eval mesh] prec = 0.0037091005431182937 (28.0/7549.0 | rec = 0.05063291139240506(28.0/553) | ap = 0.00946969696969697
[eval mesh] chair
[eval mesh] prec = 0.01814809908597165 (137.0/7549.0 | rec = 0.1253430924062214(137.0/1093) | ap = 0.10131491817235834
[eval mesh] bookshelf
[eval mesh] prec = 0.002119486024639025 (16.0/7549.0 | rec = 0.07547169811320754(16.0/212) | ap = 0.09090909090909091
[eval mesh] sofa
[eval mesh] prec = 0.007948072592396344 (60.0/7549.0 | rec = 0.5309734513274337(60.0/113) | ap = 0.416168487597059
[eval mesh] trash_bin
[eval mesh] prec = 0.010994833752814943 (83.0/7549.0 | rec = 0.3577586206896552(83.0/232) | ap = 0.18000805806512327
[eval mesh] cabinet
[eval mesh] prec = 0.017618227579811897 (133.0/7549.0 | rec = 0.5115384615384615(133.0/260) | ap = 0.26358882912551806
[eval mesh] display
[eval mesh] prec = 0.008610411975096039 (65.0/7549.0 | rec = 0.3403141361256545(65.0/191) | ap = 0.23137496193523358
[eval mesh] bathtub
[eval mesh] prec = 0.005961054444297258 (45.0/7549.0 | rec = 0.375(45.0/120) | ap = 0.15889753331566212

Is this expected? Or should I use a higher confidence threshold?

How do I use a custom dataset? What data files do I need to prepare？And what formats of point cloud data should be prepared ?

Hi,
I've been trying this code on Windows with Python3.7  Pytorch1.7.1. And I'm having this now:

subprocess.CalledProcessError: Command 'cmd /u /c "C:\Program Files (x86)\Microsoft Visual Studio\2019\BuildTools\VC\Auxiliary\Build\vcvarsall.bat" x86_amd64 && set' returned non-zero exit status 255.

The above exception was the direct cause of the following exception:
distutils.errors.DistutilsPlatformError: Microsoft Visual C++ 14.0 or greater is required. Get it with "Microsoft C++ Build Tools": https://visualstudio.microsoft.com/visu
al-cpp-build-tools/

I've installed the build tools from the given link and configure my environment variables and all possible solutions from the Internet. But the issue stayed.

Any advice?

Many thanks!

By chance, I find the data splitting files for validation and testing are identical. I wonder whether it uses the same data for validation and testing.
test data
validation data

Begin to finetune from the existing weight.
Loading checkpoint from out/pretrained_models/pretrained_weight.pth.
set() subnet missed.
Weights for finetuning loaded.
Loading data.
Traceback (most recent call last):
File "main.py", line 38, in
demo.run(cfg)
File "/home//code/RfDNet-main/demo.py", line 409, in run
our_data = generate(cfg, net.module, input_data, post_processing=False)
File "/home//code/RfDNet-main/demo.py", line 223, in generate
eval_dict, parsed_predictions = parse_predictions(end_points, data, cfg.eval_config)
File "/home/**/code/RfDNet-main/net_utils/ap_helper.py", line 257, in parse_predictions
assert (len(pick) > 0)
AssertionError

Hello, Thanks for your great work!!

I think that links for the both following datasets,

• Preprocess ScanNet and Scan2CAD data
• Preprocess ShapeNet data
  have expired again.

Can you check above datasets' link?

The demo file is not running successfully at all. It's not showing any error and logs. I tried to decipher the code and found that in main.py file its not able to import "net_utils.utils" module. The code is stuck on line 22. I tried to print a string before the import and after the import. But only a string before the import statement is printed. For reference screenshot is attached.
. I have followed every step mentioned in Readme.md file. Please help.

Hi,
thank you for making your codebase public. Unfortunately I cannot build the project, as for the environemt.yml gives an error for the pointnet. I tried to intstall it seperately but still it does not work. Are you sure the project is working with pytorch 1.7.1?

When I run command python main.py --config configs/config_files/ISCNet_detection.yaml --mode train, I met the following error.

Traceback (most recent call last):
  File "main.py", line 31, in <module>
    import train
  File "/home/Projects/RfDNet/train.py", line 4, in <module>
    from models.optimizers import load_optimizer, load_scheduler, load_bnm_scheduler
  File "/home/Projects/RfDNet/models/optimizers.py", line 5, in <module>
    from external.pointnet2.pytorch_utils import BNMomentumScheduler
ModuleNotFoundError: No module named 'external.pointnet2'

Thanks for your work!
Processed ShapeNet_data's link had been invalid ,could you please upload it again.

In your Preprocessed ScanNet data, ShapeNetv2_data/watertight_scaled is not provided (only watertight_scaled_simplified is provided). Please provide ShapeNetv2_data/watertight_scaled, thanks

The following files/folders have not been downloaded

1. ShapeNetv2_data/point.tar.gz
2. ShapeNetv2_data/pointcloud.tar.gz

How can I download the data mentioned above?

Dear @yinyunie ,
When running main.py file (after finishing all the above installations), i met an error about the ChamferDistance file.
I also run the chamfer_distance.py file alone but still met that error:

nvcc fatal : Unknown option '-generate-dependencies-with-compile'
ninja: build stopped: subcommand failed.