Update dataset card with task category and paper/code links

README.md

@@ -1,12 +1,23 @@
 ---
 license: apache-2.0
+task_categories:
+- image-to-text
+language:
+- en
+tags:
+- image-to-code
+- vision-language-models
+- code-generation
+- benchmark
 ---
 
 # Omni-I2C: Image2Code_Full
 
+[**Paper**](https://huggingface.co/papers/2603.17508) | [**GitHub**](https://github.com/MiliLab/Omni-I2C)
+
 ## Dataset Description
 
-`Image2Code_Full.tsv` is the inference split of the **Omni-I2C** benchmark. It is designed to evaluate whether multimodal models can generate high-fidelity code or structured outputs from input images.
+`Image2Code_Full.tsv` is the inference split of the **Omni-I2C** benchmark. It is designed to evaluate whether Large Multimodal Models (LMMs) can generate high-fidelity code or structured outputs from input images.
 
 Each sample contains an image, an instruction, and metadata describing the target task. The goal is to generate code or a structured string that can reproduce the original figure as accurately as possible.
 
@@ -15,7 +26,7 @@ Each sample contains an image, an instruction, and metadata describing the targe
 - **Number of figure types:** 45
 - **Number of code types:** 5
 
-Project repository: [Omni-I2C on GitHub](https://github.com/MiliLab/Omni-I2C)
+Omni-I2C requires a holistic understanding where any minor perceptual hallucination or coding error leads to a complete failure in visual reconstruction.
 
 ## Data Fields
 
@@ -23,7 +34,7 @@ The dataset is stored as a 7-column TSV file with the following header:
 
 ```tsv
 index	image	question	answer	subject	figure_type	code_type
-````
+```
 
 | Field         | Description                                                                     |
 | ------------- | ------------------------------------------------------------------------------- |
@@ -76,7 +87,7 @@ Physics
 
 ## Figure Types
 
-This split includes 45 figure types, including but not limited to:
+This split includes 45 figure types, including:
 
 ```text
 3d-plot, Area, Contour, Density, Graph, Histogram, Phase-Diagram,
@@ -94,11 +105,20 @@ scatter-plot, schematic, solid-geometry, tables, venn-diagram
 
 Within the Omni-I2C project, this file is used as the inference input to the evaluation pipeline:
 
-1. `VLMEvalKit_infer` loads `Image2Code_Full.tsv`
-2. The model takes `image` and `question` as input
-3. Predictions are saved after inference
-4. `eval_pipeline` matches predictions with GT files for code-level and image-level evaluation
+1. `VLMEvalKit_infer` loads `Image2Code_Full.tsv`.
+2. The model takes `image` and `question` as input.
+3. Predictions are saved after inference.
+4. `eval_pipeline` matches predictions with GT files for code-level and image-level evaluation.
 
 For implementation details, please refer to the project repository: [https://github.com/MiliLab/Omni-I2C](https://github.com/MiliLab/Omni-I2C)
 
+## Citation
 
+```bibtex
+@article{zhou2025omni,
+  title={Omni-I2C: A Holistic Benchmark for High-Fidelity Image-to-Code Generation},
+  author={Zhou, Jiawei and Zhang, Chi and Feng, Xiang and Zhang, Qiming and Qiu, Haibo and He, Lihuo and Ye, Dengpan and Gao, Xinbo and Zhang, Jing},
+  journal={arXiv preprint arXiv:2603.17508},
+  year={2025}
+}
+```