Add paper link, code link, and task category

This PR improves the dataset card by adding:
- The `image-to-video` task category to the YAML metadata.
- A link to the associated paper: "HighSync: High-Quality Lip Synchronization via Latent Diffusion Models".
- A link to the official GitHub repository for HighSync.
- Improved formatting for the information regarding the raw and cleaned versions of the VFHQ dataset.

README.md

@@ -1,9 +1,20 @@
-# 1-RAW VERSION
+---
+task_categories:
+- image-to-video
+---
+
+# HighSync: VFHQ Dataset
+
+This repository contains preprocessed versions of the VFHQ dataset used for the paper [HighSync: High-Quality Lip Synchronization via Latent Diffusion Models](https://huggingface.co/papers/2605.16918).
+
+**Official Code**: [https://github.com/saeed5959/high_sync](https://github.com/saeed5959/high_sync)
+
+## 1-RAW VERSION
 ### VFHQ dataset for Head-talking and Lip-syncing
 
-this is the dataset which contains raw videos of person talking in front.
+This is the dataset which contains raw videos of person talking in front.
 
-#### THe remian parts are available in :
+#### The remaining parts are available in:
 
   1-https://huggingface.co/datasets/saeed-5959/vfhq_2
   
@@ -20,7 +31,7 @@ this is the dataset which contains raw videos of person talking in front.
   7-https://huggingface.co/datasets/saeed-5959/vfhq_8
 
 ******************************************
-# 2-CLEANED VERSION
+## 2-CLEANED VERSION
 Also I have another version which I cleaned them and did preprocessing to make it ready for training of lip sync models.
 
 ### Clean pipeline :