Upload APRS dataset with Parquet format

APRS_dataset.py

@@ -0,0 +1,121 @@
+"""APRS Dataset loading script"""
+
+import csv
+import os
+import datasets
+from datasets import Features, Value, Image
+
+_CITATION = """\
+@article{aprs2024,
+  title={APRS: Allocentric and Egocentric Perspective Referring and Localization Dataset},
+  author={Your Name et al.},
+  journal={arXiv preprint},
+  year={2024}
+}
+"""
+
+_DESCRIPTION = """\
+APRS (Allocentric and egocentric Perspective Referring and localization dataSet) is a visual grounding
+dataset that contains images with both allocentric (ALLO) and egocentric (EGO) perspective descriptions
+for object localization.
+"""
+
+_HOMEPAGE = "https://huggingface.co/datasets/FudanCVL/APRS_dataset"
+_LICENSE = "MIT"
+
+_URLS = {
+    "train": ["train.csv", "train/"],
+    "test": ["test.csv", "test/"],
+}
+
+
+class APRSDataset(datasets.GeneratorBasedBuilder):
+    """APRS Dataset for visual grounding with allocentric and egocentric perspectives."""
+
+    VERSION = datasets.Version("1.0.0")
+
+    def _info(self):
+        features = Features({
+            "image": Image(),
+            "filename": Value("string"),
+            "img_width": Value("int32"),
+            "img_height": Value("int32"),
+            "aspect_ratio": Value("float32"),
+            "point_x": Value("float32"),
+            "point_y": Value("float32"),
+            "point_x_norm": Value("float32"),
+            "point_y_norm": Value("float32"),
+            "box_x": Value("float32"),
+            "box_y": Value("float32"),
+            "box_w": Value("float32"),
+            "box_h": Value("float32"),
+            "box_x_norm": Value("float32"),
+            "box_y_norm": Value("float32"),
+            "box_w_norm": Value("float32"),
+            "box_h_norm": Value("float32"),
+            "point_theta": Value("float32"),
+            "point_phi": Value("float32"),
+            "box_theta": Value("float32"),
+            "box_phi": Value("float32"),
+            "description": Value("string"),
+            "category": Value("string"),
+        })
+
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=features,
+            homepage=_HOMEPAGE,
+            license=_LICENSE,
+            citation=_CITATION,
+        )
+
+    def _split_generators(self, dl_manager):
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                gen_kwargs={
+                    "csv_path": "train.csv",
+                    "images_dir": "train",
+                },
+            ),
+            datasets.SplitGenerator(
+                name=datasets.Split.TEST,
+                gen_kwargs={
+                    "csv_path": "test.csv",
+                    "images_dir": "test",
+                },
+            ),
+        ]
+
+    def _generate_examples(self, csv_path, images_dir):
+        """Yields examples."""
+        with open(csv_path, encoding="utf-8") as f:
+            reader = csv.DictReader(f)
+            for idx, row in enumerate(reader):
+                image_path = os.path.join(images_dir, row["Filename"])
+
+                yield idx, {
+                    "image": image_path,
+                    "filename": row["Filename"],
+                    "img_width": int(float(row["Img_W"])),
+                    "img_height": int(float(row["Img_H"])),
+                    "aspect_ratio": float(row["Aspect_Ratio"]),
+                    "point_x": float(row["Pt_X"]),
+                    "point_y": float(row["Pt_Y"]),
+                    "point_x_norm": float(row["Pt_X_Norm"]),
+                    "point_y_norm": float(row["Pt_Y_Norm"]),
+                    "box_x": float(row["Box_X"]),
+                    "box_y": float(row["Box_Y"]),
+                    "box_w": float(row["Box_W"]),
+                    "box_h": float(row["Box_H"]),
+                    "box_x_norm": float(row["Box_X_Norm"]),
+                    "box_y_norm": float(row["Box_Y_Norm"]),
+                    "box_w_norm": float(row["Box_W_Norm"]),
+                    "box_h_norm": float(row["Box_H_Norm"]),
+                    "point_theta": float(row["Pt_Theta"]),
+                    "point_phi": float(row["Pt_Phi"]),
+                    "box_theta": float(row["Box_Theta"]),
+                    "box_phi": float(row["Box_Phi"]),
+                    "description": row["Description"],
+                    "category": row["Category"],
+                }

README.md

@@ -0,0 +1,139 @@
+---
+license: mit
+task_categories:
+- image-segmentation
+- visual-question-answering
+language:
+- en
+tags:
+- panoramic-vision
+- active-perception
+- referring-segmentation
+- 360-degree
+- spatial-reasoning
+size_categories:
+- 1K<n<10K
+pretty_name: Active Panoramic Referring Segmentation
+---
+
+# APRS Dataset
+
+[![Paper](https://img.shields.io/badge/arXiv-2607.02497-b31b1b.svg)](https://arxiv.org/abs/2607.02497v1)
+[![Project Page](https://img.shields.io/badge/Project-Page-green)](https://henghuiding.com/APRS/)
+[![GitHub](https://img.shields.io/badge/GitHub-Code-black)](https://github.com/FudanCVL/APRS)
+
+## Dataset Description
+
+**APRS (Active Panoramic Referring Segmentation)** is a large-scale benchmark dataset for active perception in 360° panoramic environments. Unlike passive referring segmentation that processes static images, APRS requires agents to **actively explore** continuous panoramic scenes by adjusting viewing directions to seek and segment targets based on natural language instructions.
+
+### Dataset Summary
+
+- 🎯 **7,420 samples** across **4,971 diverse panoramic scenes**
+- 🏠 Indoor and outdoor 360° environments
+- 📐 **Four types of spatial referring expressions**:
+  - **Egocentric**: First-person directional references (e.g., "look left to find...")
+  - **Unique-Attribute**: Distinctive object features (e.g., "the red sofa")
+  - **Allocentric**: Third-person spatial relations (e.g., "the chair near the window")
+  - **Multi-hop**: Complex relational reasoning (e.g., "look to the left to find the bed, then find the lamp on the table next to it")
+
+
+## Data Fields
+
+Each sample contains:
+
+- `Filename`: Image filename (panoramic view)
+- `Img_W`, `Img_H`, `Aspect_Ratio`: Image dimensions and aspect ratio
+- `Pt_X`, `Pt_Y`: Target point pixel coordinates
+- `Pt_X_Norm`, `Pt_Y_Norm`: Normalized point coordinates [0, 1]
+- `Box_X`, `Box_Y`, `Box_W`, `Box_H`: Bounding box (x, y, width, height) in pixels
+- `Box_X_Norm`, `Box_Y_Norm`, `Box_W_Norm`, `Box_H_Norm`: Normalized bounding box [0, 1]
+- `Pt_Theta`, `Pt_Phi`: Target point spherical coordinates (degrees)
+  - Theta (θ): Horizontal angle [-180°, 180°]
+  - Phi (φ): Vertical angle [-90°, 90°]
+- `Box_Theta`, `Box_Phi`: Bounding box center spherical coordinates (degrees)
+- `Description`: Natural language referring expression
+- `Category`: Spatial reference type
+  - `EGO`: Egocentric (first-person directional)
+  - `ALLO`: Allocentric (third-person spatial relations)
+  - `UNIQ`: Unique attributes
+  - `MULT`: Multi-hop reasoning
+
+## Usage
+
+### Load with Hugging Face Datasets
+
+```python
+from datasets import load_dataset
+
+# Load the dataset
+dataset = load_dataset("FudanCVL/APRS_dataset")
+
+# Access train/test splits
+train_data = dataset["train"]
+test_data = dataset["test"]
+
+# Iterate over samples
+for sample in train_data:
+    image = sample["image"]
+    description = sample["description"]
+    bbox = (sample["box_x"], sample["box_y"], sample["box_w"], sample["box_h"])
+    category = sample["category"]
+```
+
+### Load with Pandas
+
+```python
+import pandas as pd
+from PIL import Image
+
+# Load annotations
+train_df = pd.read_parquet("hf://datasets/FudanCVL/APRS_dataset/data/train.parquet")
+test_df = pd.read_parquet("hf://datasets/FudanCVL/APRS_dataset/data/test.parquet")
+
+# Access a sample
+sample = train_df.iloc[0]
+print(f"Description: {sample['Description']}")
+print(f"Category: {sample['Category']}")
+print(f"Spherical coords: θ={sample['Pt_Theta']:.2f°, φ={sample['Pt_Phi']:.2f°}")
+```
+
+### Use with PanoSeeker Agent
+
+```bash
+# Clone the official repository
+git clone https://github.com/FudanCVL/APRS.git
+cd APRS
+
+# Install dependencies
+pip install -e ".[all]"
+
+# Download the dataset
+# (Place in APRS_dataset/ directory)
+
+# Visualize panoramas
+aprs-viewer --root APRS_dataset --split test --index 0
+```
+
+## Citation
+
+If you use this dataset, please cite:
+
+```bibtex
+@article{tang2026seek,
+  title={Seek to Segment: Active Perception for Panoramic Referring Segmentation},
+  author={Tang, Song and Hu, Shuming and Shuai, Xincheng and Ding, Henghui and Jiang, Yu-Gang},
+  journal={arXiv preprint arXiv:2607.02497},
+  year={2026}
+}
+```
+
+## License
+
+This dataset is released under the [MIT License](https://opensource.org/licenses/MIT).
+
+## Links
+
+- 📄 **Paper**: [arXiv:2607.02497](https://arxiv.org/abs/2607.02497v1)
+- 🌐 **Project Page**: [https://henghuiding.com/APRS/](https://henghuiding.com/APRS/)
+- 💻 **Code**: [https://github.com/FudanCVL/APRS](https://github.com/FudanCVL/APRS)
+

data/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bfc10d28621584c5f68bc7b42add8d95951dcdf0985f0b10f3bbc98cec44468
+size 317639

data/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:640c7bf574827a4bb4c3b8bb1d25d8801223a643cff90eb4c2f6477ef577a2c8
+size 760847