ScanNet Example¶

This example trains a semantic segmentation model on ScanNet indoor scenes using a MinkUNet-style encoder-decoder built with sparse convolutions.

Dataset¶

The script uses the pre-processed ScanNet 3D point clouds from the OpenScene project. Each scene is stored as (coords, colors, labels):

• coords: (N, 3) float32 — 3D point positions
• colors: (N, 3) float32 — RGB color features
• labels: (N,) int — semantic class labels (20 classes, 255 = ignore)

The 20 semantic classes include: wall, floor, cabinet, bed, chair, sofa, table, door, window, bookshelf, picture, counter, desk, curtain, refrigerator, shower curtain, toilet, sink, bathtub, and other furniture.

The dataset is downloaded automatically on first run (~1.3 GB) to ./data/scannet_3d/.

This example does not apply augmentations (random rotation, scaling, color jitter, etc.). For high-quality training results, implement your own augmentation pipeline.

Network architecture¶

The default model is MinkUNet18, a U-Net with sparse convolution encoder and decoder blocks connected by skip connections. Available models:

Model	Description
mink_unet.MinkUNet18	Lightweight U-Net (default)
mink_unet.MinkUNet34	Deeper encoder
mink_unet.MinkUNet50	ResNet-50 style blocks
mink_unet.MinkUNet101	ResNet-101 style blocks

Input points are voxelized at voxel_size=0.02 and wrapped via PointToSparseWrapper, which handles the point-to-voxel conversion and maps output features back to the original point resolution.

The model outputs per-point logits with shape (N, 20).

Setup¶

Install the optional model and training dependencies:

pip install "warpconvnet[models]"

Additional requirements: hydra-core, omegaconf, torchmetrics.

Run¶

python examples/scannet.py

The script uses Hydra for configuration. Override any parameter on the command line:

# Smaller batch size for limited GPU memory
python examples/scannet.py train.batch_size=4

# Use a deeper model
python examples/scannet.py model._target_=mink_unet.MinkUNet34

# Change voxel size and learning rate
python examples/scannet.py data.voxel_size=0.05 train.lr=0.01

Configuration reference¶

Paths:

Key	Default	Description
paths.data_dir	./data/scannet_3d	Dataset directory
paths.output_dir	./results/	Output directory
paths.ckpt_path	null	Checkpoint path to resume from

Training:

Key	Default	Description
train.batch_size	12	Training batch size
train.lr	0.001	AdamW learning rate
train.epochs	100	Number of training epochs
train.step_size	20	StepLR decay period (epochs)
train.gamma	0.7	StepLR decay factor
train.num_workers	8	DataLoader workers

Test:

Key	Default	Description
test.batch_size	12	Test batch size
test.num_workers	4	DataLoader workers

Data:

Key	Default	Description
data.num_classes	20	Number of semantic classes
data.voxel_size	0.02	Voxelization resolution (meters)
data.ignore_index	255	Label index to ignore in loss/metrics

Model:

Key	Default	Description
model._target_	mink_unet.MinkUNet18	Model class to instantiate
model.in_channels	3	Input feature channels (RGB)
model.out_channels	20	Output channels (num classes)
model.in_type	voxel	Input type (voxel wraps model with PointToSparseWrapper)

General:

Key	Default	Description
device	cuda	Device
use_wandb	false	Enable Weights & Biases logging
seed	42	Random seed

Expected output¶

Each epoch prints a progress bar followed by test-set evaluation with accuracy and mean IoU:

Train Epoch: 1 Loss:  2.143: 100%|██████████| 104/104
Test set: Average loss:  1.8234, Accuracy:  42.15%, mIoU:  18.73%

After 100 epochs with default settings, expect roughly:

• Overall accuracy: ~75-80%
• mIoU: ~55-65%

Results will vary with augmentation, model choice, and voxel size. This example is intended as a starting point, not a benchmark-tuned recipe.