README.md

---
viewer: false
license: other
license_name: shapenet-license
license_link: https://shapenet.org/terms
task_categories:
  - other
tags:
  - 3d
  - sdf
  - signed-distance-field
  - shapenet
  - point-cloud
  - neural-fields
  - implicit-neural-representation
pretty_name: ShapeNetSDF
size_categories:
  - 100B<n<1T
configs:
  - config_name: airplane
    data_files:
      - split: train
        path: data/airplane/train-*.parquet
      - split: val
        path: data/airplane/val-*.parquet
      - split: test
        path: data/airplane/test-*.parquet
  - config_name: bag
    data_files:
      - split: train
        path: data/bag/train-*.parquet
      - split: val
        path: data/bag/val-*.parquet
      - split: test
        path: data/bag/test-*.parquet
  - config_name: basket
    data_files:
      - split: train
        path: data/basket/train-*.parquet
      - split: val
        path: data/basket/val-*.parquet
      - split: test
        path: data/basket/test-*.parquet
  - config_name: bathtub
    data_files:
      - split: train
        path: data/bathtub/train-*.parquet
      - split: val
        path: data/bathtub/val-*.parquet
      - split: test
        path: data/bathtub/test-*.parquet
  - config_name: bed
    data_files:
      - split: train
        path: data/bed/train-*.parquet
      - split: val
        path: data/bed/val-*.parquet
      - split: test
        path: data/bed/test-*.parquet
  - config_name: bench
    data_files:
      - split: train
        path: data/bench/train-*.parquet
      - split: val
        path: data/bench/val-*.parquet
      - split: test
        path: data/bench/test-*.parquet
  - config_name: birdhouse
    data_files:
      - split: train
        path: data/birdhouse/train-*.parquet
      - split: val
        path: data/birdhouse/val-*.parquet
      - split: test
        path: data/birdhouse/test-*.parquet
  - config_name: bookshelf
    data_files:
      - split: train
        path: data/bookshelf/train-*.parquet
      - split: val
        path: data/bookshelf/val-*.parquet
      - split: test
        path: data/bookshelf/test-*.parquet
  - config_name: bottle
    data_files:
      - split: train
        path: data/bottle/train-*.parquet
      - split: val
        path: data/bottle/val-*.parquet
      - split: test
        path: data/bottle/test-*.parquet
  - config_name: bowl
    data_files:
      - split: train
        path: data/bowl/train-*.parquet
      - split: val
        path: data/bowl/val-*.parquet
      - split: test
        path: data/bowl/test-*.parquet
  - config_name: bus
    data_files:
      - split: train
        path: data/bus/train-*.parquet
      - split: val
        path: data/bus/val-*.parquet
      - split: test
        path: data/bus/test-*.parquet
  - config_name: cabinet
    data_files:
      - split: train
        path: data/cabinet/train-*.parquet
      - split: val
        path: data/cabinet/val-*.parquet
      - split: test
        path: data/cabinet/test-*.parquet
  - config_name: camera
    data_files:
      - split: train
        path: data/camera/train-*.parquet
      - split: val
        path: data/camera/val-*.parquet
      - split: test
        path: data/camera/test-*.parquet
  - config_name: can
    data_files:
      - split: train
        path: data/can/train-*.parquet
      - split: val
        path: data/can/val-*.parquet
      - split: test
        path: data/can/test-*.parquet
  - config_name: cap
    data_files:
      - split: train
        path: data/cap/train-*.parquet
      - split: val
        path: data/cap/val-*.parquet
      - split: test
        path: data/cap/test-*.parquet
  - config_name: car
    data_files:
      - split: train
        path: data/car/train-*.parquet
      - split: val
        path: data/car/val-*.parquet
      - split: test
        path: data/car/test-*.parquet
  - config_name: chair
    data_files:
      - split: train
        path: data/chair/train-*.parquet
      - split: val
        path: data/chair/val-*.parquet
      - split: test
        path: data/chair/test-*.parquet
  - config_name: dishwasher
    data_files:
      - split: train
        path: data/dishwasher/train-*.parquet
      - split: val
        path: data/dishwasher/val-*.parquet
      - split: test
        path: data/dishwasher/test-*.parquet
  - config_name: display
    data_files:
      - split: train
        path: data/display/train-*.parquet
      - split: val
        path: data/display/val-*.parquet
      - split: test
        path: data/display/test-*.parquet
  - config_name: earphone
    data_files:
      - split: train
        path: data/earphone/train-*.parquet
      - split: val
        path: data/earphone/val-*.parquet
      - split: test
        path: data/earphone/test-*.parquet
  - config_name: faucet
    data_files:
      - split: train
        path: data/faucet/train-*.parquet
      - split: val
        path: data/faucet/val-*.parquet
      - split: test
        path: data/faucet/test-*.parquet
  - config_name: file_cabinet
    data_files:
      - split: train
        path: data/file_cabinet/train-*.parquet
      - split: val
        path: data/file_cabinet/val-*.parquet
      - split: test
        path: data/file_cabinet/test-*.parquet
  - config_name: guitar
    data_files:
      - split: train
        path: data/guitar/train-*.parquet
      - split: val
        path: data/guitar/val-*.parquet
      - split: test
        path: data/guitar/test-*.parquet
  - config_name: jar
    data_files:
      - split: train
        path: data/jar/train-*.parquet
      - split: val
        path: data/jar/val-*.parquet
      - split: test
        path: data/jar/test-*.parquet
  - config_name: keyboard
    data_files:
      - split: train
        path: data/keyboard/train-*.parquet
      - split: val
        path: data/keyboard/val-*.parquet
      - split: test
        path: data/keyboard/test-*.parquet
  - config_name: knife
    data_files:
      - split: train
        path: data/knife/train-*.parquet
      - split: val
        path: data/knife/val-*.parquet
      - split: test
        path: data/knife/test-*.parquet
  - config_name: lamp
    data_files:
      - split: train
        path: data/lamp/train-*.parquet
      - split: val
        path: data/lamp/val-*.parquet
      - split: test
        path: data/lamp/test-*.parquet
  - config_name: laptop
    data_files:
      - split: train
        path: data/laptop/train-*.parquet
      - split: val
        path: data/laptop/val-*.parquet
      - split: test
        path: data/laptop/test-*.parquet
  - config_name: loudspeaker
    data_files:
      - split: train
        path: data/loudspeaker/train-*.parquet
      - split: val
        path: data/loudspeaker/val-*.parquet
      - split: test
        path: data/loudspeaker/test-*.parquet
  - config_name: mailbox
    data_files:
      - split: train
        path: data/mailbox/train-*.parquet
      - split: val
        path: data/mailbox/val-*.parquet
      - split: test
        path: data/mailbox/test-*.parquet
  - config_name: microphone
    data_files:
      - split: train
        path: data/microphone/train-*.parquet
      - split: val
        path: data/microphone/val-*.parquet
      - split: test
        path: data/microphone/test-*.parquet
  - config_name: motorbike
    data_files:
      - split: train
        path: data/motorbike/train-*.parquet
      - split: val
        path: data/motorbike/val-*.parquet
      - split: test
        path: data/motorbike/test-*.parquet
  - config_name: mug
    data_files:
      - split: train
        path: data/mug/train-*.parquet
      - split: val
        path: data/mug/val-*.parquet
      - split: test
        path: data/mug/test-*.parquet
  - config_name: piano
    data_files:
      - split: train
        path: data/piano/train-*.parquet
      - split: val
        path: data/piano/val-*.parquet
      - split: test
        path: data/piano/test-*.parquet
  - config_name: pillow
    data_files:
      - split: train
        path: data/pillow/train-*.parquet
      - split: val
        path: data/pillow/val-*.parquet
      - split: test
        path: data/pillow/test-*.parquet
  - config_name: pistol
    data_files:
      - split: train
        path: data/pistol/train-*.parquet
      - split: val
        path: data/pistol/val-*.parquet
      - split: test
        path: data/pistol/test-*.parquet
  - config_name: printer
    data_files:
      - split: train
        path: data/printer/train-*.parquet
      - split: val
        path: data/printer/val-*.parquet
      - split: test
        path: data/printer/test-*.parquet
  - config_name: remote
    data_files:
      - split: train
        path: data/remote/train-*.parquet
      - split: val
        path: data/remote/val-*.parquet
      - split: test
        path: data/remote/test-*.parquet
  - config_name: rifle
    data_files:
      - split: train
        path: data/rifle/train-*.parquet
      - split: val
        path: data/rifle/val-*.parquet
      - split: test
        path: data/rifle/test-*.parquet
  - config_name: rocket
    data_files:
      - split: train
        path: data/rocket/train-*.parquet
      - split: val
        path: data/rocket/val-*.parquet
      - split: test
        path: data/rocket/test-*.parquet
  - config_name: skateboard
    data_files:
      - split: train
        path: data/skateboard/train-*.parquet
      - split: val
        path: data/skateboard/val-*.parquet
      - split: test
        path: data/skateboard/test-*.parquet
  - config_name: sofa
    data_files:
      - split: train
        path: data/sofa/train-*.parquet
      - split: val
        path: data/sofa/val-*.parquet
      - split: test
        path: data/sofa/test-*.parquet
  - config_name: stove
    data_files:
      - split: train
        path: data/stove/train-*.parquet
      - split: val
        path: data/stove/val-*.parquet
      - split: test
        path: data/stove/test-*.parquet
  - config_name: table
    data_files:
      - split: train
        path: data/table/train-*.parquet
      - split: val
        path: data/table/val-*.parquet
      - split: test
        path: data/table/test-*.parquet
  - config_name: telephone
    data_files:
      - split: train
        path: data/telephone/train-*.parquet
      - split: val
        path: data/telephone/val-*.parquet
      - split: test
        path: data/telephone/test-*.parquet
  - config_name: tower
    data_files:
      - split: train
        path: data/tower/train-*.parquet
      - split: val
        path: data/tower/val-*.parquet
      - split: test
        path: data/tower/test-*.parquet
  - config_name: train
    data_files:
      - split: train
        path: data/train/train-*.parquet
      - split: val
        path: data/train/val-*.parquet
      - split: test
        path: data/train/test-*.parquet
  - config_name: trash_bin
    data_files:
      - split: train
        path: data/trash_bin/train-*.parquet
      - split: val
        path: data/trash_bin/val-*.parquet
      - split: test
        path: data/trash_bin/test-*.parquet
  - config_name: washer
    data_files:
      - split: train
        path: data/washer/train-*.parquet
      - split: val
        path: data/washer/val-*.parquet
      - split: test
        path: data/washer/test-*.parquet
  - config_name: watercraft
    data_files:
      - split: train
        path: data/watercraft/train-*.parquet
      - split: val
        path: data/watercraft/val-*.parquet
      - split: test
        path: data/watercraft/test-*.parquet
  - config_name: all
    data_files:
      - split: train
        path: data/*/train-*.parquet
      - split: val
        path: data/*/val-*.parquet
      - split: test
        path: data/*/test-*.parquet
---

# ShapeNetSDF

Signed Distance Field (SDF) point samples derived from
[ShapeNet Core](#), for training and evaluating implicit
neural representations / neural fields on 3D shapes.

This dataset is shared as part of the CVPR 2026 paper [Weight Space Representation Learning via Neural Field Adaptaion](https://arxiv.org/abs/2512.01759). 
Code for producing this dataset is shared in the [wsr.pytorch](https://github.com/IVRL/wsr.pytorch) neural-field codebase.

<img src="https://inrainbws.github.io/assets/images/wsrl.png" width="640">

Each shape is converted into a watertight manifold, normalized into the unit
cube `[-1, 1]³`, and sampled into three point sets (`uniform`, `surface`,
`groundtruth`), each stored as a `[N, 4]` `float32` array of `[x, y, z, sdf]`.

## Dataset structure

The dataset is stored as **sharded Parquet** (one config per category, with
`train` / `val` / `test` splits). Each row is one shape; the three point sets
(`uniform`, `surface`, `groundtruth`) are stored as fixed-shape `[262144, 4]`
float32 tensors in their own columns.

```
ShapeNetSDF/
├── data/                           # heavy data: sharded parquet, one folder per category
│   └── <category>/                 # 50 category configs (chair, table, airplane, ...)
│       ├── train-NNNNN-of-NNNNN.parquet   # rows: model_id, uniform, surface, groundtruth
│       ├── val-NNNNN-of-NNNNN.parquet
│       └── test-NNNNN-of-NNNNN.parquet
├── meta/                           # lightweight metadata (split id lists, labels, subsets)
│   ├── <category>/                 # per-category split id lists
│   │   └── train.txt / val.txt / test.txt
│   └── all/                        # global splits + curated subsets + labels
│       ├── train.txt / val.txt / test.txt   # global splits (all categories)
│       ├── 10k10c.txt / 5k10c.txt / 5k5c.txt / 100_5c.txt   # curated subsets (model-id lists)
│       └── labels.json             # category ↔ model_id mappings
└── README.md
```

- Each Parquet **row** has columns `model_id` (string) and `uniform` /
  `surface` / `groundtruth`, each an `Array2D[262144, 4]` float32 tensor
  (`[x, y, z, sdf]` per point — see *File format* below).
- The `all` config globs every category's Parquet, so
  `load_dataset(..., "all")` streams all 50 categories together. The split id
  lists under `meta/` (and curated subsets) remain for selecting specific model ids.

- **50 categories**: `airplane, bag, basket, bathtub, bed, bench, birdhouse,
  bookshelf, bottle, bowl, bus, cabinet, camera, can, cap, car, chair,
  dishwasher, display, earphone, faucet, file_cabinet, guitar, jar, keyboard,
  knife, lamp, laptop, loudspeaker, mailbox, microphone, motorbike, mug, piano,
  pillow, pistol, printer, remote, rifle, rocket, skateboard, sofa, stove,
  table, telephone, tower, train, trash_bin, washer, watercraft` (plus the
  aggregated `all/` folder).
- **Total size**: ~512 GB.

### File format

Each point set (`uniform` / `surface` / `groundtruth`) is a `float32` array of
shape `[262144, 4]`:

| column | meaning | range |
| ------ | ------- | ----- |
| 0 (`x`) | x coordinate | `[-1, 1]` |
| 1 (`y`) | y coordinate | `[-1, 1]` |
| 2 (`z`) | z coordinate | `[-1, 1]` |
| 3 (`sdf`) | signed distance to the surface | negative inside, positive outside |

- **`uniform`** — `N = 64³ = 262,144` points sampled uniformly in `[-1, 1]³`,
  with their exact signed distance. Use these to supervise the field in free
  space.
- **`surface`** — points sampled on the mesh surface with a small Gaussian
  perturbation (noise scale ≈ `0.02`), so SDF values are close to (but not
  exactly) zero. Intended for training near the surface.
- **`groundtruth`** — exact on-surface points (SDF ≈ `0`), with no noise.
  Intended for evaluation (e.g. surface reconstruction / Chamfer / IoU).

### Splits

Each category config has `train` / `val` / `test` splits (~80% / 10% / 10%,
seed `42`), encoded as the Parquet partitions. Plain-text `model_id` lists under
`meta/` remain for global splits and curated subsets:

- `meta/all/train.txt` / `val.txt` / `test.txt` — global splits over all categories
  (per-category lists live in `meta/<category>/`).
- `meta/all/{10k10c,5k10c,5k5c,100_5c}.txt` — curated subsets
  (`<num_models><num_categories>c`, e.g. `5k5c` = 5,000 models across 5
  categories) for quick experiments. Filter the loaded dataset by `model_id`.
- `meta/all/labels.json` — `{"category_to_filename": {...}, "filename_to_category": {...}}`
  mapping each category to its model ids and vice versa.

## Usage

```python
from datasets import load_dataset

# One category, one split. Array2D columns decode to numpy [262144, 4] arrays.
ds = load_dataset("EPFL-IVRL/ShapeNetSDF", "chair", split="train").with_format("numpy")
row = ds[0]
xyz, sdf = row["uniform"][:, :3], row["uniform"][:, 3]   # [262144, 3], [262144]
# row also has "surface" and "groundtruth", same shape, plus "model_id".

# All categories together (streamed to avoid downloading everything at once):
ds_all = load_dataset("EPFL-IVRL/ShapeNetSDF", "all", split="train", streaming=True)
```

Or download the Parquet files for a category locally:

```python
from huggingface_hub import snapshot_download

snapshot_download(
    repo_id="EPFL-IVRL/ShapeNetSDF",
    repo_type="dataset",
    allow_patterns=["data/chair/**", "meta/all/labels.json"],
    local_dir="ShapeNetSDF",
)
```

## How it was created

For each ShapeNet Core model (`models/model_normalized.obj`):

1. Convert the mesh to a watertight manifold (via `point_cloud_utils`).
2. Normalize into the unit cube centered at the origin (scale `0.98` of the max
   vertex distance).
3. Sample `uniform`, `surface` (+noise), and `groundtruth` point sets.
4. Compute the signed distance for every point and save as `[N, 4]` `float32`.

Processing is deterministic (per-`model_id` seeding), so results are
reproducible. See `neural_field/scripts/process_shapenet_to_sdf.py` for the full
pipeline and configuration.

## License & citation

This dataset is created as part of the CVPR 2026 Paper "Weight Space Representation Learning
via Neural Field Adaptation".

```bibtex
@inproceedings{yang2026wsr,
  title     = {Weight Space Representation Learning via Neural Field Adaptation},
  author    = {Yang, Zhuoqian and Salzmann, Mathieu and S{\"u}sstrunk, Sabine},
  booktitle = {Proceedings of the IEEE/CVF Conference on
               Computer Vision and Pattern Recognition (CVPR)},
  year      = {2026}
}
```

This dataset is derived from **ShapeNet** and is subject to the
[ShapeNet terms of use](https://shapenet.org/terms). Use it for non-commercial
research only, and cite ShapeNet:

```bibtex
@article{chang2015shapenet,
  title   = {ShapeNet: An Information-Rich 3D Model Repository},
  author  = {Chang, Angel X. and Funkhouser, Thomas and Guibas, Leonidas and
             Hanrahan, Pat and Huang, Qixing and Li, Zimo and Savarese, Silvio
             and Savva, Manolis and Song, Shuran and Su, Hao and Xiao, Jianxiong
             and Yi, Li and Yu, Fisher},
  journal = {arXiv preprint arXiv:1512.03012},
  year    = {2015}
}
```