causal3d

.gitattributes

@@ -17,6 +17,7 @@
 *.npz filter=lfs diff=lfs merge=lfs -text
 *.onnx filter=lfs diff=lfs merge=lfs -text
 *.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
 *.pb filter=lfs diff=lfs merge=lfs -text
 *.pickle filter=lfs diff=lfs merge=lfs -text
 *.pkl filter=lfs diff=lfs merge=lfs -text
@@ -56,6 +57,3 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
-dataset_metadata.json filter=lfs diff=lfs merge=lfs -text
-*.parquet filter=lfs diff=lfs merge=lfs -text
-parquet/*.parquet filter=lfs diff=lfs merge=lfs -text

Causal3D.py

@@ -1,184 +0,0 @@
-import datasets
-import pandas as pd
-import os
-from pathlib import Path
-from tqdm import tqdm
-
-_CITATION = """\
-@article{liu2025causal3d,
-  title={CAUSAL3D: A Comprehensive Benchmark for Causal Learning from Visual Data},
-  author={Liu, Disheng and Qiao, Yiran and Liu, Wuche and Lu, Yiren and Zhou, Yunlai and Liang, Tuo and Yin, Yu and Ma, Jing},
-  journal={arXiv preprint arXiv:2503.04852},
-  year={2025}
-}
-"""
-
-_DESCRIPTION = """\
-Causal3D is a benchmark for evaluating causal reasoning in physical and hypothetical visual scenes. 
-It includes both real-world recordings and rendered synthetic scenes demonstrating causal interactions.
-"""
-
-_HOMEPAGE = "https://huggingface.co/datasets/LLDDSS/Causal3D"
-_LICENSE = "CC-BY-4.0"
-
-class Causal3D(datasets.GeneratorBasedBuilder):
-    DEFAULT_CONFIG_NAME = "real_scenes_Water_flow_scene_render"
-    BUILDER_CONFIGS = [
-        # hypothetical_scenes
-        datasets.BuilderConfig(name="hypothetical_scenes_Hypothetic_v2_linear", version=datasets.Version("1.0.0"), description="Hypothetic_v2_linear scene"),
-        datasets.BuilderConfig(name="hypothetical_scenes_Hypothetic_v2_nonlinear", version=datasets.Version("1.0.0"), description="Hypothetic_v2_nonlinear scene"),
-        datasets.BuilderConfig(name="hypothetical_scenes_Hypothetic_v3_fully_connected_linear", version=datasets.Version("1.0.0"), description="Hypothetic_v3_fully_connected_linear scene"),
-        datasets.BuilderConfig(name="hypothetical_scenes_Hypothetic_v4_linear_full_connected", version=datasets.Version("1.0.0"), description="Hypothetic_v4_linear_full_connected scene"),
-        datasets.BuilderConfig(name="hypothetical_scenes_Hypothetic_v4_linear_v", version=datasets.Version("1.0.0"), description="Hypothetic_v4_linear_v scene"),
-        datasets.BuilderConfig(name="hypothetical_scenes_Hypothetic_v4_nonlinear_v", version=datasets.Version("1.0.0"), description="Hypothetic_v4_nonlinear_v scene"),
-        datasets.BuilderConfig(name="hypothetical_scenes_Hypothetic_v5_linear", version=datasets.Version("1.0.0"), description="Hypothetic_v5_linear scene"),
-        datasets.BuilderConfig(name="hypothetical_scenes_Hypothetic_v5_linear_full_connected", version=datasets.Version("1.0.0"), description="Hypothetic_v5_linear_full_connected scene"),
-        datasets.BuilderConfig(name="hypothetical_scenes_rendered_h3_linear_128P", version=datasets.Version("1.0.0"), description="rendered_h3_linear_128P scene"),
-        datasets.BuilderConfig(name="hypothetical_scenes_rendered_h3_nonlinear_128P", version=datasets.Version("1.0.0"), description="rendered_h3_nonlinear_128P scene"),
-        datasets.BuilderConfig(name="hypothetical_scenes_rendered_h5_nonlinear", version=datasets.Version("1.0.0"), description="rendered_h5_nonlinear scene"),
-
-        # real_scenes
-        datasets.BuilderConfig(name="real_scenes_Real_Parabola", version=datasets.Version("1.0.0"), description="Real_Parabola scene"),
-        datasets.BuilderConfig(name="real_scenes_Real_magnet_v3", version=datasets.Version("1.0.0"), description="Real_magnet_v3 scene"),
-        datasets.BuilderConfig(name="real_scenes_Real_magnet_v3_5", version=datasets.Version("1.0.0"), description="Real_magnet_v3_5 scene"),
-        # datasets.BuilderConfig(name="real_scenes_Real_Parabola_multi_view", version=datasets.Version("1.0.0"), description="Real_parabola_multi_view scene"),
-        datasets.BuilderConfig(name="real_scenes_Real_spring_v3_256P", version=datasets.Version("1.0.0"), description="Real_spring_v3_256P scene"),
-        datasets.BuilderConfig(name="real_scenes_Water_flow_scene_render", version=datasets.Version("1.0.0"), description="Water_flow_scene_render scene"),
-        datasets.BuilderConfig(name="real_scenes_convex_len_render_images", version=datasets.Version("1.0.0"), description="convex_len_render_images scene"),
-        datasets.BuilderConfig(name="real_scenes_real_pendulum", version=datasets.Version("1.0.0"), description="real_pendulum scene"),
-        datasets.BuilderConfig(name="real_scenes_rendered_magnetic_128", version=datasets.Version("1.0.0"), description="rendered_magnetic_128 scene"),
-        datasets.BuilderConfig(name="real_scenes_rendered_reflection_128P", version=datasets.Version("1.0.0"), description="rendered_reflection_128P scene"),
-        datasets.BuilderConfig(name="real_scenes_seesaw_scene_128P", version=datasets.Version("1.0.0"), description="seesaw_scene_128P scene"),
-        datasets.BuilderConfig(name="real_scenes_spring_scene_128P", version=datasets.Version("1.0.0"), description="spring_scene_128P scene"),
-    ]
-
-    def _info(self):
-        return datasets.DatasetInfo(
-            description=_DESCRIPTION,
-            features=datasets.Features({
-                "image": datasets.Image(),
-                "file_name": datasets.Value("string"),
-                "metadata": datasets.Value("string"),  # optionally replace with structured fields
-            }),
-            homepage=_HOMEPAGE,
-            license=_LICENSE,
-            citation=_CITATION,
-        )
-    
-    # def _split_generators(self, dl_manager):
-    #     parts = self.config.name.split("_", 2)
-        
-    #     category = parts[0] + "_" + parts[1]  # real_scenes or hypothetical_scenes
-
-    #     if category not in ["real_scenes", "hypothetical_scenes"]:
-    #         raise ValueError(f"Invalid category '{category}'. Must be one of ['real_scenes', 'hypothetical_scenes']")
-
-    #     scene = parts[2]
-    #     data_dir = os.path.join(category, scene)
-
-    #     return [
-    #         datasets.SplitGenerator(
-    #             name=datasets.Split.TRAIN,
-    #             gen_kwargs={"data_dir": data_dir},
-    #         )
-    #     ]
-    def _split_generators(self, dl_manager):
-        parts = self.config.name.split("_", 2)
-        category = parts[0] + "_" + parts[1]
-        if category not in ["real_scenes", "hypothetical_scenes"]:
-            raise ValueError(f"Invalid category '{category}'.")
-
-        scene = parts[2]
-        scene_path = os.path.join(category, scene)
-
-        if os.path.exists(scene_path):  
-          data_dir = scene_path
-        else: 
-          archive_path = dl_manager.download_and_extract(f"{scene_path}.zip")
-          data_dir = os.path.join(archive_path, scene)
-
-        return [
-            datasets.SplitGenerator(
-                name=datasets.Split.TRAIN,
-                gen_kwargs={"data_dir": data_dir},
-            )
-        ]
-    def _generate_examples(self, data_dir):
-        def color(text, code):
-            return f"\033[{code}m{text}\033[0m"
-        print("load data from {}".format(data_dir))  
-        try:
-            image_files = {}
-            for ext in ("*.png", "*.jpg", "*.jpeg"):
-                for img_path in Path(data_dir).rglob(ext):
-                    relative_path = str(img_path.relative_to(data_dir))
-                    image_files[relative_path] = str(img_path)
-            parts = [i.split('/')[0] for i in list(image_files.keys())]
-            parts = set(parts)
-            if "part_000" not in parts:
-               parts= ['']
-
-
-        except Exception as e:
-            print(color(f"Error loading images: {e}", "31"))  # Red
-            return
-    
-        # Find the .csv file
-        csv_files = list(Path(data_dir).rglob("*.csv"))
-        csv_files = [f for f in Path(data_dir).rglob("*.csv") if not f.name.startswith("._")]
-        if not csv_files:
-            # print(f"\033[33m[SKIP] No CSV found in {data_dir}, skipping this config.\033[0m")
-            pass
-        # print(f"\033[33m[INFO] Found CSV: {csv_files}\033[0m")
-        csv_path = csv_files[0] if csv_files else None
-        df = pd.read_csv(csv_path) if csv_path else None
-        image_col_exists = True
-        if df is not None and "image" not in df.columns:
-            image_col_exists = False
-                
-        images = df["image"].tolist() if image_col_exists and df is not None else []
-        images = [i.split('/')[-1].split('.')[0] for i in images if i.endswith(('.png', '.jpg', '.jpeg'))]
-
-        try:
-            # Match CSV rows with image paths
-            if df is None:
-                for i, j in tqdm(image_files.items(), desc="Processing images", unit="image"):
-                    yield i, {
-                        "image": j,
-                        "file_name": i,
-                        "metadata": None,
-                    }
-
-            else:
-                for idx, row in tqdm(df.iterrows(), total=len(df), desc="Processing rows", unit="row"):
-                    fname = row["ID"]
-                    raw_record_img_path = images[idx] if images else "" #row["image"] 
-                    record_img_name = raw_record_img_path.split('/')[-1]
-                    for part in parts:
-                        if part == '':
-                            record_img_path = record_img_name
-                        else:
-                            record_img_path = "/".join([part, record_img_name.strip()])
-                            if "Water_flow_scene_render" in data_dir:
-                                record_img_path = "/".join([part, str(int(record_img_name.strip().split('.')[0]))+".png"])
-                        if record_img_path in image_files:
-                            # print(color(f"record_img_path: { image_files[record_img_path]}", "34"))  # Blue
-                            yield idx, {
-                                "image": image_files[record_img_path],
-                                "file_name": fname,
-                                "metadata": row.to_json(),
-                            }
-                            break
-                            
-                        else:
-                            yield idx, {
-                                # "image": "",
-                                "file_name": fname,
-                                "metadata": row.to_json(),
-                            }
-                            break
-
-
-        except Exception as e:
-            print(color(f"Error processing CSV rows: {e}", "31"))
-        

README.md

@@ -1,419 +1,10 @@
 ---
 license: cc-by-4.0
-language:
-- en
-size_categories:
-- 100K<n<1M
-pretty_name: Causal3D
-tags:
-- Causality
-- Computer_Vision
-dataset_info:
-- config_name: hypothetical_scenes_Hypothetic_v2_linear
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 2197142
-    num_examples: 14368
-  download_size: 0
-  dataset_size: 2197142
-- config_name: hypothetical_scenes_Hypothetic_v2_nonlinear
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 1809956
-    num_examples: 10000
-  download_size: 0
-  dataset_size: 1809956
-- config_name: hypothetical_scenes_Hypothetic_v3_fully_connected_linear
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 1397093
-    num_examples: 10000
-  download_size: 0
-  dataset_size: 1397093
-- config_name: hypothetical_scenes_Hypothetic_v4_linear_full_connected
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 1699598
-    num_examples: 10050
-  download_size: 0
-  dataset_size: 1699598
-- config_name: hypothetical_scenes_Hypothetic_v4_linear_v
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 2053379
-    num_examples: 10000
-  download_size: 0
-  dataset_size: 2053379
-- config_name: hypothetical_scenes_Hypothetic_v4_nonlinear_v
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 2828217
-    num_examples: 10000
-  download_size: 0
-  dataset_size: 2828217
-- config_name: hypothetical_scenes_Hypothetic_v5_linear
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 1956461
-    num_examples: 10000
-  download_size: 0
-  dataset_size: 1956461
-- config_name: hypothetical_scenes_Hypothetic_v5_linear_full_connected
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 1955921
-    num_examples: 10000
-  download_size: 0
-  dataset_size: 1955921
-- config_name: hypothetical_scenes_rendered_h3_linear_128P
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 5425498
-    num_examples: 15000
-  download_size: 0
-  dataset_size: 5425498
-- config_name: hypothetical_scenes_rendered_h3_nonlinear_128P
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 3239120
-    num_examples: 10223
-  download_size: 0
-  dataset_size: 3239120
-- config_name: hypothetical_scenes_rendered_h5_nonlinear
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 5459126
-    num_examples: 10360
-  download_size: 0
-  dataset_size: 5459126
-- config_name: real_scenes_Real_Parabola
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 1323548
-    num_examples: 10000
-  download_size: 0
-  dataset_size: 1323548
-- config_name: real_scenes_Real_magnet_v3
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 28397
-    num_examples: 481
-  download_size: 0
-  dataset_size: 28397
-- config_name: real_scenes_Real_magnet_v3_5
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 93977
-    num_examples: 1503
-  download_size: 0
-  dataset_size: 93977
-- config_name: real_scenes_Real_parabola_multi_view
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 0
-    num_examples: 0
-  download_size: 0
-  dataset_size: 0
-- config_name: real_scenes_Real_spring_v3_256P
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 136325
-    num_examples: 450
-  download_size: 0
-  dataset_size: 136325
-- config_name: real_scenes_Water_flow_scene_render
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 2792618
-    num_examples: 10000
-  download_size: 0
-  dataset_size: 2792618
-- config_name: real_scenes_convex_len_render_images
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 72448
-    num_examples: 1078
-  download_size: 0
-  dataset_size: 72448
-- config_name: real_scenes_real_pendulum
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 2925963
-    num_examples: 9999
-  download_size: 0
-  dataset_size: 2925963
-- config_name: real_scenes_rendered_magnetic_128
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 2324526
-    num_examples: 8350
-  download_size: 0
-  dataset_size: 2324526
-- config_name: real_scenes_rendered_reflection_128P
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 2765222
-    num_examples: 9995
-  download_size: 0
-  dataset_size: 2765222
-- config_name: real_scenes_seesaw_scene_128P
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 2275814
-    num_examples: 10000
-  download_size: 0
-  dataset_size: 2275814
-- config_name: real_scenes_spring_scene_128P
-  features:
-  - name: image
-    dtype: image
-  - name: file_name
-    dtype: string
-  - name: metadata
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 2547386
-    num_examples: 10000
-  download_size: 0
-  dataset_size: 2547386
 ---
 # 🧠 Causal3D: A Benchmark for Visual Causal Reasoning
 
 **Causal3D** is a comprehensive benchmark designed to evaluate models’ abilities to uncover *latent causal relations* from structured and visual data. This dataset integrates **3D-rendered scenes** with **tabular causal annotations**, providing a unified testbed for advancing *causal discovery*, *causal representation learning*, and *causal reasoning* with **vision-language models (VLMs)** and **large language models (LLMs)**.
 
-## 🖼️ Visual Previews
-
-Below are example images from different Causal3D scenes:
-
-<table>
-  <tr>
-    <td align="center">
-      <img src="https://huggingface.co/datasets/LLDDSS/Causal3D/resolve/main/preview/parabola.png" width="250"/><br/>parabola
-    </td>
-    <td align="center">
-      <img src="https://huggingface.co/datasets/LLDDSS/Causal3D/resolve/main/preview/convex.png" width="250"/><br/>convex
-    </td>
-  </tr>
-  <tr>
-    <td align="center">
-      <img src="https://huggingface.co/datasets/LLDDSS/Causal3D/resolve/main/preview/magnetic.png" width="200"/><br/>magnetic
-    </td>
-    <td align="center">
-      <img src="https://huggingface.co/datasets/LLDDSS/Causal3D/resolve/main/preview/pendulum.png" width="200"/><br/>pendulum
-    </td>
-    <td align="center">
-      <img src="https://huggingface.co/datasets/LLDDSS/Causal3D/resolve/main/preview/reflection.png" width="200"/><br/>reflection
-    </td>
-  </tr>
-  <tr>
-    <td align="center">
-      <img src="https://huggingface.co/datasets/LLDDSS/Causal3D/resolve/main/preview/seesaw.png" width="200"/><br/>seesaw
-    </td>
-    <td align="center">
-      <img src="https://huggingface.co/datasets/LLDDSS/Causal3D/resolve/main/preview/spring.png" width="200"/><br/>spring
-    </td>
-    <td align="center">
-      <img src="https://huggingface.co/datasets/LLDDSS/Causal3D/resolve/main/preview/water_flow.png" width="200"/><br/>water_flow
-    </td>
-  </tr>
-</table>
-
-<!-- - `causal_graph.json`: Ground-truth causal structure (as adjacency matrix or graph).
-- `view_info.json`: Camera/viewpoint metadata.
-- `split.json`: Recommended train/val/test splits for benchmarking. -->
-
-
-## 📚 Usage
-
-#### 🔹 Option 1: Load from Hugging Face
-
-You can easily load a specific scene using the Hugging Face `datasets` library:
-
-```python
-from datasets import load_dataset
-
-dataset = load_dataset(
-    "LLDDSS/Causal3D",
-    name="real_scenes_Real_Parabola",
-    download_mode="force_redownload",  # Optional: force re-download
-    trust_remote_code=True             # Required for custom dataset loading
-)
-
-print(dataset)
-```
-
-#### 🔹 Option 2: Download via [**Kaggle**](https://www.kaggle.com/datasets/dsliu0011/causal3d-image-dataset) + Croissant
-```python
-import mlcroissant as mlc
-import pandas as pd
-
-# Load the dataset metadata from Kaggle
-croissant_dataset = mlc.Dataset(
-    "https://www.kaggle.com/datasets/dsliu0011/causal3d-image-dataset/croissant/download"
-)
-
-record_sets = croissant_dataset.metadata.record_sets
-print(record_sets)
-
-df = pd.DataFrame(croissant_dataset.records(record_set=record_sets[0].uuid))
-print(df.head())
-```
-
 ---
 
 ## 📌 Overview
@@ -430,9 +21,11 @@ While recent progress in AI and computer vision has been remarkable, there remai
 
 Each sub-dataset (scene) contains:
 
-
 - `images/`: Rendered images under different camera views and backgrounds.
-- `tabular.csv`: Instance-level annotations including object attributes in causal graph.
+- `metadata.csv`: Instance-level annotations including object attributes and positions.
+<!-- - `causal_graph.json`: Ground-truth causal structure (as adjacency matrix or graph).
+- `view_info.json`: Camera/viewpoint metadata.
+- `split.json`: Recommended train/val/test splits for benchmarking. -->
 
 ---
 
@@ -462,8 +55,6 @@ We evaluate a diverse set of methods:
 
 ---
 
-
-
 <!-- ## 🔍 Example Use Case
 
 ```python

__init__.py

@@ -1 +0,0 @@
-from .Causal3D_ import Causal3D