Update README.md

README.md

@@ -12,146 +12,124 @@ tags:
 - visual-grounding
 ---
 
-# Dataset Card for Dataset Name
-
-<!-- Provide a quick summary of the dataset. -->
-
-
+# Dataset Card for Multi-Agent Referential Communication Dataset
 
 ## Dataset Details
 
 ### Dataset Description
 
-<!-- Provide a longer summary of what this dataset is. -->
+This dataset contains spatial dialogue data for multi-agent referential communication tasks in 3D environments. It includes pairs of images showing speaker and listener views within photorealistic indoor scenes, along with natural language descriptions of target object locations.
 
-
-This dataset contains spatial dialogue data for visual grounding tasks. It includes pairs of images showing speaker and listener views, along with natural language descriptions of spatial locations and target object positions.
+The key feature of this dataset is that it captures communication between two agents with different physical perspectives in a shared 3D space. Each agent has their own unique viewpoint of the scene, requiring them to consider each other's perspectives when generating and interpreting spatial references.
 
 ### Dataset Summary
 
-The dataset consists of 2,970 dialogue instances where speakers provide spatial descriptions of target objects, and listeners attempt to identify the correct object based on these descriptions.
-
-
-- **Curated by:** [More Information Needed]
-- **Funded by [optional]:** [More Information Needed]
-- **Shared by [optional]:** [More Information Needed]
-- **Language(s) (NLP):** en
-- **License:** apache-2.0
-
-### Dataset Sources [optional]
-
-<!-- Provide the basic links for the dataset. -->
-
-- **Repository:** [More Information Needed]
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
+- **Size**: 2,970 dialogue instances across 1,485 scenes
+- **Total Scenes Generated**: 27,504 scenes (24,644 train, 1,485 validation, 1,375 test)
+- **Task Type**: Referential communication between embodied agents
+- **Language(s)**: English
+- **License**: Apache-2.0
+- **Curated by**: University of California, Berkeley
+- **Time per Task**: Median 33.0s for speakers, 10.5s for listeners
 
 ## Uses
 
-<!-- Address questions around how the dataset is intended to be used. -->
-
 ### Direct Use
 
-<!-- This section describes suitable use cases for the dataset. -->
-
-[More Information Needed]
+The dataset is designed for:
+- Training and evaluating referring expression generation models
+- Training and evaluating visual question answering systems
+- Studying human spatial language use in multi-perspective scenarios
+- Developing embodied AI systems that can communicate about shared environments
+- Research on perspective-taking in language generation and comprehension
 
 ### Out-of-Scope Use
 
-<!-- This section addresses misuse, malicious use, and uses that the dataset will not work well for. -->
-
-[More Information Needed]
+The dataset should not be used for:
+- Training systems to navigate or manipulate physical environments
+- Training general-purpose vision-language models without consideration of perspective
+- Applications requiring real-time interaction or dialogue (dataset contains single-turn interactions only)
 
 ## Dataset Structure
 
-<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->
-
-[More Information Needed]
+Each instance contains:
+- Speaker view image (1280x720 resolution)
+- Listener view image (1280x720 resolution)
+- Natural language referring expression from speaker
+- Target object location
+- Listener object selection
+- Scene metadata including:
+  - Agent positions and orientations
+  - Field of view overlap measurements
+  - Referent placement method (random vs adversarial)
+  - Base environment identifier
 
 ## Dataset Creation
 
 ### Curation Rationale
 
-<!-- Motivation for the creation of this dataset. -->
-
-[More Information Needed]
+The dataset was created to study how humans and AI systems handle referential communication when embodied in different physical perspectives within a shared environment. The multi-perspective nature of the task reflects real-world scenarios where agents must coordinate despite having different viewpoints.
 
 ### Source Data
 
-<!-- This section describes the source data (e.g. news text and headlines, social media posts, translated sentences, ...). -->
-
 #### Data Collection and Processing
 
-<!-- This section describes the data collection and processing process such as data selection criteria, filtering and normalization methods, tools and libraries used, etc. -->
-
-[More Information Needed]
+1. Base environments from ScanNet++ (450 high-quality 3D indoor environments)
+2. Scene generation process:
+   - Place two agents with controlled relative orientations (0° to 180°)
+   - Place 3 referent objects using either random or adversarial placement
+   - Render images from each agent's perspective
+   - Apply quality filtering using GPT-4V
+3. Human data collection:
+   - 194 qualified crowdworkers on Prolific
+   - Speakers describe target object location
+   - Listeners select object based on description
+   - 3 listener judgments per description
 
 #### Who are the source data producers?
 
-<!-- This section describes the people or systems who originally created the data. It should also include self-reported demographic or identity information for the source data creators if this information is available. -->
-
-[More Information Needed]
-
-### Annotations [optional]
-
-<!-- If the dataset contains annotations which are not part of the initial data collection, use this section to describe them. -->
-
-#### Annotation process
+- Base 3D environments: ScanNet++ dataset
+- Referring expressions: English-speaking crowdworkers from the United States
+- Quality filtering: Automated GPT-4V system
+- Scene generation: Automated system with physics simulation
 
-<!-- This section describes the annotation process such as annotation tools used in the process, the amount of data annotated, annotation guidelines provided to the annotators, interannotator statistics, annotation validation, etc. -->
+### Personal and Sensitive Information
 
-[More Information Needed]
-
-#### Who are the annotators?
-
-<!-- This section describes the people or systems who created the annotations. -->
-
-[More Information Needed]
-
-#### Personal and Sensitive Information
-
-<!-- State whether the dataset contains data that might be considered personal, sensitive, or private (e.g., data that reveals addresses, uniquely identifiable names or aliases, racial or ethnic origins, sexual orientations, religious beliefs, political opinions, financial or health data, etc.). If efforts were made to anonymize the data, describe the anonymization process. -->
-
-[More Information Needed]
+The dataset does not contain personally identifiable information. Crowdworker data was checked to exclude private information and offensive content.
 
 ## Bias, Risks, and Limitations
 
-<!-- This section is meant to convey both technical and sociotechnical limitations. -->
-
-[More Information Needed]
+- Limited to indoor environments from ScanNet++
+- English language only
+- Single-turn interactions only (no dialogue)
+- Restricted to specific object types (spheres)
+- May reflect cultural biases in spatial language use
+- Limited demographic diversity of crowdworkers
 
 ### Recommendations
 
-<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
-
-Users should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.
+- Consider cultural and linguistic differences in spatial language when using the dataset
+- Account for perspective differences when developing models
+- Evaluate performance across different relative orientations and referent placements
+- Consider expanding to multi-turn dialogue in future work
+- Test for biases in spatial language use across different demographics
 
-## Citation [optional]
-
-<!-- If there is a paper or blog post introducing the dataset, the APA and Bibtex information for that should go in this section. -->
+## Citation
 
 **BibTeX:**
+```
+@article{tang2024grounding,
+  title={Grounding Language in Multi-Perspective Referential Communication},
+  author={Tang, Zineng and Mao, Lingjun and Suhr, Alane},
+  journal={EMNLP},
+  year={2024}
+}
+```
 
-[More Information Needed]
-
-**APA:**
-
-[More Information Needed]
-
-## Glossary [optional]
-
-<!-- If relevant, include terms and calculations in this section that can help readers understand the dataset or dataset card. -->
-
-[More Information Needed]
-
-## More Information [optional]
-
-[More Information Needed]
-
-## Dataset Card Authors [optional]
+## Dataset Card Contact
 
-[More Information Needed]
+Contact the authors at {terran, lingjun, suhr}@berkeley.edu
 
-## Dataset Card Contact
+## More Information
 
-[More Information Needed]
+Code, models, and dataset available at: https://github.com/zinengtang/MulAgentRef