Upload README.md with huggingface_hub

README.md

@@ -14,278 +14,94 @@ tags:
 
 # Objaverse VIDA Dataset
 
-This dataset contains processed 3D assets from [Objaverse](https://objaverse.allenai.org/) prepared for use with [AI2-THOR](https://ai2thor.allenai.org/) and [ProcTHOR](https://procthor.allenai.org/) environments. It includes ~40,000 processed 3D objects with textures and metadata, plus ~160,000 procedurally generated house layouts.
+Processed 3D assets from [Objaverse](https://objaverse.allenai.org/) for use with [AI2-THOR](https://ai2thor.allenai.org/) and [ProcTHOR](https://procthor.allenai.org/).
 
-## What's Included
+## Contents
 
 | Component | Description | Size |
 |-----------|-------------|------|
-| `processed_2023_07_28/` | 39,664 processed 3D objects from Objaverse with textures and AI2-THOR metadata | 27 GB |
-| `houses_2023_07_28/` | 160,000+ procedurally generated house layouts (train/test/val splits) | 3.7 GB |
-| `procthor_databases_2023_07_28/` | Asset databases, material definitions, placement annotations | 70 MB |
-| `0.json` | Sample house layout | 90 KB |
+| `processed_2023_07_28/` | ~40K processed 3D objects with textures | 27 GB |
+| `houses_2023_07_28/` | 160K+ procedurally generated house layouts | 3.7 GB |
+| `procthor_databases_2023_07_28/` | Asset databases, materials, placement rules | 70 MB |
 
-## Data Format
-
-### Processed Objects (`processed_2023_07_28/`)
-
-Each object is stored in its own directory named by UUID. Each directory contains:
-
-```
-{object_id}/
-├── {object_id}.pkl.gz    # Processed 3D mesh data (gzip-compressed pickle)
-├── albedo.jpg            # Albedo/diffuse texture map
-├── normal.jpg            # Normal map  
-├── emission.jpg          # Emission map
-└── thor_metadata.json    # AI2-THOR compatible metadata (bounding box, properties, etc.)
-```
-
-### House Layouts (`houses_2023_07_28/`)
-
-- `train.jsonl.gz` / `test.jsonl.gz` / `val.jsonl.gz` - Compressed JSONL files with full house definitions
-- `train/` / `test/` / `val/` directories - Individual house JSON files (packed in `.tar` archives)
-
-Each house JSON contains rooms, doors, windows, objects, walls, and procedural parameters for AI2-THOR.
-
-### Asset Databases (`procthor_databases_2023_07_28/`)
-
-- `asset-database.json` - Full asset catalog with metadata
-- `material-database.json` - Material definitions
-- `placement-annotations.json` - Object placement rules
-- `receptacles.json` - Receptacle definitions for object placement
-- `asset_groups/` - Predefined furniture groupings
-
----
-
-## Unpacking Instructions
-
-The data is archived in `.tar` files for efficient storage. **To restore the original directory structure**, follow these steps:
-
-### Quick Start (Full Dataset)
+## Quick Start
 
 ```bash
-# 1. Install huggingface_hub with hf_transfer for fast downloads
+# Install dependencies
 pip install huggingface_hub[hf_transfer]
-export HF_HUB_ENABLE_HF_TRANSFER=1
-
-# 2. Download and unpack everything
-python -c "
-from huggingface_hub import snapshot_download
-snapshot_download(
-    repo_id='spatial-training/objaverse_vida',
-    repo_type='dataset',
-    local_dir='./objaverse_vida'
-)
-"
-
-# 3. Unpack the tar archives to restore original structure
-cd objaverse_vida
-
-# Unpack processed objects (13 shards → ~40K directories)
-mkdir -p processed_2023_07_28_unpacked
-for shard in processed_2023_07_28/shard_*.tar; do
-    echo "Extracting $shard..."
-    tar -xf "$shard" -C processed_2023_07_28_unpacked/
-done
-# Replace archived version with unpacked
-rm -rf processed_2023_07_28/*.tar
-mv processed_2023_07_28_unpacked/* processed_2023_07_28/
-rmdir processed_2023_07_28_unpacked
-
-# Unpack house individual files
-cd houses_2023_07_28
-for split in train test val; do
-    if [ -f "${split}_individual.tar" ]; then
-        echo "Extracting ${split}_individual.tar..."
-        mkdir -p "$split"
-        tar -xf "${split}_individual.tar" -C "$split/"
-        rm "${split}_individual.tar"
-    fi
-done
-cd ..
-
-echo "Done! Dataset restored to original structure."
-```
-
-### Python Script (Recommended)
-
-Save this as `unpack_dataset.py`:
-
-```python
-#!/usr/bin/env python3
-"""Download and unpack objaverse_vida to original directory structure."""
 
-import os
-import tarfile
-from pathlib import Path
-from huggingface_hub import snapshot_download
+# Download the unpack script
+wget https://huggingface.co/datasets/spatial-training/objaverse_vida/raw/main/unpack.py
 
-def unpack_dataset(target_dir: str = "./objaverse_vida"):
-    target = Path(target_dir)
-    
-    # Enable hf_transfer for fast downloads
-    os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"
-    
-    print("Downloading dataset...")
-    snapshot_download(
-        repo_id="spatial-training/objaverse_vida",
-        repo_type="dataset",
-        local_dir=str(target)
-    )
-    
-    # Unpack processed shards
-    processed_dir = target / "processed_2023_07_28"
-    if processed_dir.exists():
-        print("\nUnpacking processed objects...")
-        shards = sorted(processed_dir.glob("shard_*.tar"))
-        for shard in shards:
-            print(f"  Extracting {shard.name}...")
-            with tarfile.open(shard) as tar:
-                tar.extractall(processed_dir)
-            shard.unlink()  # Remove tar after extraction
-        # Remove manifest (no longer needed after extraction)
-        manifest = processed_dir / "manifest.json"
-        if manifest.exists():
-            manifest.unlink()
-    
-    # Unpack houses individual files
-    houses_dir = target / "houses_2023_07_28"
-    if houses_dir.exists():
-        print("\nUnpacking house files...")
-        for split in ["train", "test", "val"]:
-            tar_file = houses_dir / f"{split}_individual.tar"
-            if tar_file.exists():
-                print(f"  Extracting {tar_file.name}...")
-                split_dir = houses_dir / split
-                split_dir.mkdir(exist_ok=True)
-                with tarfile.open(tar_file) as tar:
-                    tar.extractall(split_dir)
-                tar_file.unlink()  # Remove tar after extraction
-    
-    print(f"\nDone! Dataset unpacked to: {target.absolute()}")
-    print("\nStructure:")
-    print("  processed_2023_07_28/     - 39,664 object directories")
-    print("  houses_2023_07_28/        - train/test/val splits")
-    print("  procthor_databases_2023_07_28/")
-    print("  0.json")
-
-if __name__ == "__main__":
-    import sys
-    target = sys.argv[1] if len(sys.argv) > 1 else "./objaverse_vida"
-    unpack_dataset(target)
+# Run it (downloads ~30GB and extracts to ./objaverse_vida)
+python unpack.py ./objaverse_vida
 ```
 
-Run with:
+Or specify a custom path:
 ```bash
-python unpack_dataset.py /path/to/destination
+python unpack.py /path/to/destination
 ```
 
-### Download Specific Components Only
-
-```python
-from huggingface_hub import hf_hub_download
-import tarfile
+## Data Format
 
-# Download just one processed shard
-shard_path = hf_hub_download(
-    repo_id="spatial-training/objaverse_vida",
-    filename="processed_2023_07_28/shard_00.tar",
-    repo_type="dataset"
-)
+### Processed Objects
 
-# Extract to local directory
-with tarfile.open(shard_path) as tar:
-    tar.extractall("./processed_2023_07_28")
+Each object directory contains:
+```
+{object_id}/
+├── {object_id}.pkl.gz    # 3D mesh data (gzip pickle)
+├── albedo.jpg            # Diffuse texture
+├── normal.jpg            # Normal map
+├── emission.jpg          # Emission map
+└── thor_metadata.json    # AI2-THOR metadata
 ```
 
-### Find Which Shard Contains a Specific Object
+### House Layouts
 
-```python
-from huggingface_hub import hf_hub_download
-import json
+- `train.jsonl.gz` / `test.jsonl.gz` / `val.jsonl.gz` - Full house definitions
+- `train/` / `test/` / `val/` - Individual house JSON files
 
-# Download manifest
-manifest_path = hf_hub_download(
-    repo_id="spatial-training/objaverse_vida",
-    filename="processed_2023_07_28/manifest.json",
-    repo_type="dataset"
-)
+### Asset Databases
 
-with open(manifest_path) as f:
-    manifest = json.load(f)
+- `asset-database.json` - Asset catalog
+- `material-database.json` - Materials
+- `placement-annotations.json` - Placement rules
+- `receptacles.json` - Receptacle definitions
 
-# Find object by ID
-object_id = "000074a334c541878360457c672b6c2e"
-for shard_name, shard_info in manifest["shards"].items():
-    if object_id in shard_info["dirs"]:
-        print(f"Object {object_id} is in {shard_name}")
-        break
-```
+## Manual Download
 
----
+If you prefer not to use the script:
 
-## Final Directory Structure (After Unpacking)
+```bash
+# Clone the dataset
+huggingface-cli download spatial-training/objaverse_vida --repo-type dataset --local-dir ./objaverse_vida
 
-```
-objaverse_vida/
-├── 0.json
-├── processed_2023_07_28/
-│   ├── 000074a334c541878360457c672b6c2e/
-│   │   ├── 000074a334c541878360457c672b6c2e.pkl.gz
-│   │   ├── albedo.jpg
-│   │   ├── emission.jpg
-│   │   ├── normal.jpg
-│   │   └── thor_metadata.json
-│   ├── 0002e50309b44e409c96f440202d90b3/
-│   │   └── ...
-│   └── ... (39,664 object directories)
-├── houses_2023_07_28/
-│   ├── train.jsonl.gz
-│   ├── test.jsonl.gz
-│   ├── val.jsonl.gz
-│   ├── train/
-│   │   ├── 0.json.gz
-│   │   ├── 1.json.gz
-│   │   └── ... (159,745 files)
-│   ├── test/
-│   │   └── ... (15,994 files)
-│   └── val/
-│       └── ... (15,829 files)
-└── procthor_databases_2023_07_28/
-    ├── asset-database.json
-    ├── material-database.json
-    ├── placement-annotations.json
-    ├── receptacles.json
-    ├── refined_annotations.json
-    ├── skyboxes.json
-    ├── solid-wall-colors.json
-    ├── wall-holes.json
-    └── asset_groups/
-        └── ... (furniture groupings)
-```
+# Extract processed shards
+cd objaverse_vida/processed_2023_07_28
+for f in shard_*.tar; do tar -xf "$f" && rm "$f"; done
+rm manifest.json
 
----
+# Extract house files
+cd ../houses_2023_07_28
+for split in train test val; do
+  tar -xf "${split}_individual.tar" -C "$split/" && rm "${split}_individual.tar"
+done
+```
 
 ## Citation
 
-If you use this dataset, please cite the original works:
-
 ```bibtex
 @inproceedings{deitke2023objaverse,
   title={Objaverse: A Universe of Annotated 3D Objects},
-  author={Deitke, Matt and Schwenk, Dustin and Salvador, Jordi and Weihs, Luca and Michel, Oscar and VanderBilt, Eli and Schmidt, Ludwig and Ehsani, Kiana and Kembhavi, Aniruddha and Farhadi, Ali},
+  author={Deitke, Matt and others},
   booktitle={CVPR},
   year={2023}
 }
 
 @inproceedings{deitke2022procthor,
   title={ProcTHOR: Large-Scale Embodied AI Using Procedural Generation},
-  author={Deitke, Matt and VanderBilt, Eli and Herrasti, Alvaro and Weihs, Luca and Ehsani, Kiana and Salvador, Jordi and Han, Winson and Kolve, Eric and Kembhavi, Aniruddha and Mottaghi, Roozbeh},
+  author={Deitke, Matt and others},
   booktitle={NeurIPS},
   year={2022}
 }
 ```
-
-## License
-
-Apache 2.0