Update README.md

README.md

@@ -26,13 +26,46 @@ size_categories:
 </div>
 
 # Spatial Perception and Reasoning Benchmark (SPBench)
-This repository contains the Spatial Perception and Reasoning Benchmark (SPBench), introduced in [SpatialLadder: Progressive Training for Spatial Reasoning in Vision-Language Models]().
-
-## Files
 
+This repository contains the Spatial Perception and Reasoning Benchmark (SPBench), introduced in [SpatialLadder: Progressive Training for Spatial Reasoning in Vision-Language Models]().
 
 ## Dataset Description
+
 SPBench is a comprehensive evaluation suite designed to assess the spatial perception and reasoning capabilities of Vision-Language Models (VLMs). SPBench consists of two complementary benchmarks, SPBench-SI and SPBench-MV, corresponding to single-image and multi-view modalities, respectively. Both benchmarks are constructed using the standardized pipeline applied to the ScanNet validation set, ensuring systematic coverage across diverse spatial reasoning tasks.
 
-SPBench-SI serves as a single-image evaluation benchmark that measures models’ ability to perform spatial understanding and reasoning from individual viewpoints. It encompasses four task categories—absolute distance, object size, relative distance, and relative direction—containing a total of 1,009 samples. In contrast, SPBench-MV focuses on multi-view spatial reasoning, requiring models to jointly model spatial relationships across multiple viewpoints. It further includes object counting tasks to evaluate models’ capability in identifying and enumerating objects within multi-view scenarios. Both benchmarks undergo rigorous quality control through a combination of standardized pipeline filtering strategies and manual curation, ensuring data disambiguation and high-quality annotations suitable for reliable evaluation.
+SPBench-SI serves as a single-image evaluation benchmark that measures models’ ability to perform spatial understanding and reasoning from individual viewpoints. It encompasses four task categories—absolute distance, object size, relative distance, and relative direction, with a total of 1,009 samples. In contrast, SPBench-MV focuses on multi-view spatial reasoning, requiring models to jointly model spatial relationships across multiple viewpoints. It further includes object counting tasks to evaluate models’ capability in identifying and enumerating objects within multi-view scenarios, with a total of 319 samples. Both benchmarks undergo rigorous quality control through a combination of standardized pipeline filtering strategies and manual curation, ensuring data disambiguation and high-quality annotations suitable for reliable evaluation.
+
+## Usage
+
+You can directly load the dataset from Hugging Face using the `datasets` library.
+SPBench can be accessed in three different configurations as follows:
+
+```python
+from datasets import load_dataset
+
+# Load the two benchmarks directly
+dataset = load_dataset("Gino319/SPBench")
+
+# Load the SPBench-SI only
+dataset = load_dataset("Gino319/SPBench", data_files="SPBench-SI.parquet")
+
+# Load the SPBench-MV only
+dataset = load_dataset("Gino319/SPBench", data_files="SPBench-MV.parquet")
+```
+
+The corresponding image resources required for the benchmarks are provided in `SPBench-SI-images.zip`
+and `SPBench-MV-images.zip`, which contain the complete image sets for SPBench-SI and SPBench-MV, respectively.
+
+## Evaluation
+
+SPBench evaluates performance using two metrics: for multiple-choice questions, we use `Accuracy`, calculated based on exact matches. For numerical questions, we use `MRA (Mean Relative Accuracy)` introduced by [Thinking in Space](https://github.com/vision-x-nyu/thinking-in-space), to assess how closely model predictions align with ground truth values.
+
+The evaluation code and usage guidelines are available in our [GitHub repository](https://github.com/ZJU-REAL/SpatialLadder). For comprehensive details, please refer to our paper and the repository documentation.
+
+## Citation
+
+```bibtex
+```
+
+