Cylinder3D

https://github.com/xinge008/cylinder3d
Admin
Cylinder3D implements Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR segmentation,...
Tokens:1,972
Snippets:6
Trust Score:7.7
License:Apache-2.0
Update:1 year ago
Show doc for...
Tokens:
Raw
### Train nuScenes Dataset

Source: https://github.com/xinge008/cylinder3d/blob/master/NUSCENES-GUIDE.md

This script is used to train a model on the nuScenes dataset. It requires a custom configuration file to be modified with user-specific paths and settings.

```Shell
#!/bin/bash

# Example command to train on nuScenes dataset
# python train_nusc.py --config /path/to/your/config.yaml
```

--------------------------------

### Prepare nuScenes Data

Source: https://github.com/xinge008/cylinder3d/blob/master/README.md

This snippet illustrates the directory structure for preparing the nuScenes dataset. It indicates the expected files and folders for the dataset.

```Shell
./
├── 
├── ...
└── path_to_data_shown_in_config/
		├──v1.0-trainval
		├──v1.0-test
		├──samples
		├──sweeps
		├──maps


```

--------------------------------

### Run Semantic Segmentation Demo

Source: https://github.com/xinge008/cylinder3d/blob/master/README.md

This Python command executes a demo for semantic segmentation on a folder of LiDAR scans. It requires specifying the input demo folder and an output save folder.

```Python
python demo_folder.py --demo-folder YOUR_FOLDER --save-folder YOUR_SAVE_FOLDER
```

--------------------------------

### Run Semantic Segmentation Demo with Labels

Source: https://github.com/xinge008/cylinder3d/blob/master/README.md

This Python command runs the semantic segmentation demo and includes an optional argument for providing ground truth labels, which is useful for validation.

```Python
python demo_folder.py --demo-folder YOUR_FOLDER --save-folder YOUR_SAVE_FOLDER --demo-label-folder YOUR_LABEL_FOLDER
```

--------------------------------

### Prepare SemanticKITTI Data

Source: https://github.com/xinge008/cylinder3d/blob/master/README.md

This snippet shows the directory structure for preparing SemanticKITTI dataset. It outlines the expected organization of velodyne scans and labels within sequence folders.

```Shell
./
├── 
├── ...
└── path_to_data_shown_in_config/
    ├──sequences
        ├── 00/           
        │   ├── velodyne/	
        |   |	├── 000000.bin
        |   |	├── 000001.bin
        |   |	└── ...
        │   └── labels/ 
        |       ├── 000000.label
        |       ├── 000001.label
        |       └── ...
        ├── 08/ # for validation
        ├── 11/ # 11-21 for testing
        └── 21/
	    └── ...

```

--------------------------------

### Train Cylinder3D Model

Source: https://github.com/xinge008/cylinder3d/blob/master/README.md

This command initiates the training process for the Cylinder3D model. It assumes that the configuration file (e.g., semantickitti.yaml) has been modified with custom settings.

```Shell
sh train.sh
```
Tokens:
Raw