Upload sonata-lp.scannet20.fair

README.md

@@ -0,0 +1,115 @@
+---
+license: cc-by-nc-4.0
+library_name: torch-pointcloud
+tags:
+- point-cloud
+- 3d
+- pytorch
+- torch-pointcloud
+- sonata
+- segmentation
+datasets:
+- scannet
+base_model: torch-pointcloud/sonata-base.pretrain.fair
+model-index:
+- name: sonata-lp.scannet20.fair
+  results:
+  - task:
+      type: point-cloud-segmentation
+    dataset:
+      name: ScanNet (20 classes)
+      type: scannet
+    metrics:
+    - name: mIoU
+      type: mean_iou
+      value: 72.6
+---
+
+# Model card for sonata-lp.scannet20.fair
+
+A Sonata point cloud segmentation model (self-distilled point representation encoder). Trained on ScanNet (20 classes).
+
+> **Non-commercial.** These weights are released by [facebookresearch/sonata](https://github.com/facebookresearch/sonata) under CC BY-NC 4.0 and may be used for research and evaluation only.
+
+## Model Details
+
+- **Model Type:** Point cloud semantic segmentation
+- **Model Stats:**
+  - Params (M): 108.5
+  - Input channels: 9
+  - Classes: 20
+  - Features: 1232
+- **Dataset:** ScanNet (20 classes)
+- **Metrics:** mIoU 72.6 (reference 72.5)
+- **Paper:** [Sonata: Self-Supervised Learning of Reliable Point Representations](https://arxiv.org/abs/2503.16429)
+- **Converted from:** [facebookresearch/sonata](https://github.com/facebookresearch/sonata) (CC-BY-NC-4.0)
+- **Library:** [torch-pointcloud](https://github.com/arthurdjn/pytorch-pointcloud)
+
+## Install
+
+```bash
+pip install torch-pointcloud
+```
+
+## Usage
+
+```python
+import torch
+import torch_pointcloud as tp
+from torch_pointcloud.utils.data import collate
+
+model, info = tp.create_model(
+    "sonata-lp.scannet20.fair",
+    task="segmentation",
+    pretrained=True,
+    return_info=True,
+)
+model = model.cuda().eval()  # GPU-only kernels
+
+# synthetic sample with the keys a dataset provides
+num_points = 8192
+sample = {
+    "pos": torch.randn(num_points, 3),
+    "color": torch.rand(num_points, 3) * 255,
+    "normal": torch.randn(num_points, 3),
+    "segment": torch.zeros(num_points, dtype=torch.long),
+    "instance": torch.zeros(num_points, dtype=torch.long),
+}
+data = info["transform"](sample)
+data = collate([data])
+data = {key: value.cuda() for key, value in data.items()}
+
+with torch.no_grad():
+    logits = model(data.get("x"), data["pos_grid"], data["batch"])
+```
+
+## Feature extraction
+
+```python
+with torch.no_grad():
+    features = model.forward_features(data.get("x"), data["pos_grid"], data["batch"])
+
+model.reset_classifier(num_classes=0)
+with torch.no_grad():
+    features = model(data.get("x"), data["pos_grid"], data["batch"])  # (N, 1232)
+```
+
+## Citation
+
+```bibtex
+@inproceedings{wu2025sonata,
+  title   = {Sonata: Self-Supervised Learning of Reliable Point Representations},
+  author  = {Xiaoyang Wu and Daniel DeTone and Duncan Frost and Tianwei Shen and Chris Xie and Nan Yang and Jakob Engel and Richard Newcombe and Hengshuang Zhao and Julian Straub},
+  booktitle = {CVPR},
+  year    = {2025}
+}
+```
+
+```bibtex
+@inproceedings{dai2017scannet,
+  title   = {ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes},
+  author  = {Angela Dai and Angel X. Chang and Manolis Savva and Maciej Halber and Thomas Funkhouser and Matthias Nießner},
+  booktitle = {CVPR},
+  year    = {2017}
+}
+```

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b284c23f2258381d2f54d751cdc59e2f6ee675ccb9dd79ad1ae3df6f1d3ea87
+size 433993176