Improve model card: add pipeline tag, links, and clean metadata

README.md

@@ -3,9 +3,10 @@ license: other
 license_name: license
 license_link: LICENSE
 spaces:
-  - viciopoli/REALM-demo
-arxiv: 2605.00271
+- viciopoli/REALM-demo
+pipeline_tag: image-feature-extraction
 ---
+
 <a name="readme-top"></a>
 
 <div align="center" style="background-color: #0e2841; padding: 20px; border-radius: 15px; margin-bottom: 20px;">
@@ -15,6 +16,7 @@ arxiv: 2605.00271
 
 Welcome to **REALM**! This repository contains the implementation of REALM, for advanced computer vision tasks involving both traditional RGB and Event-based vision.
 
+[📄 Paper](https://huggingface.co/papers/2605.00271) | [💻 GitHub](https://github.com/utiasSTARS/REALM) | [🌐 Project Page](https://papers.starslab.ca/realm/) | [🤗 Demo](https://huggingface.co/spaces/viciopoli/REALM-demo)
 
 <div align="center">
     <img src="media/demo_realm.gif" alt="demo" >
@@ -68,7 +70,7 @@ Event cameras provide several unique advantages over standard frame-based sensor
 
 We address this gap with **REALM**, a cross-modal framework that learns an **R**GB and **E**vent **A**ligned **L**atent **M**anifold by projecting event representations into the pretrained latent space of RGB foundation models. Instead of task-specific training, we leverage low-rank adaptation (LoRA) to bridge the modality gap, effectively unlocking the geometric and semantic priors of frozen RGB backbones for asynchronous event streams.
 
-We demonstrate that **REALM** effectively maps events into the ViT-based foundation latent space. Our method allows us to perform downstream tasks like depth estimation and semantic segmentation by simply transferring linear heads trained on the RGB teacher. Most significantly, **REALM** enables the direct, zero-shot application of complex, frozen image-trained decoders, such as MASt3R, to raw event data. We demonstrate state-of-the-art performance in wide-baseline feature matching, significantly outperforming specialized architectures. Code and models are available upon acceptance.
+We demonstrate that **REALM** effectively maps events into the ViT-based foundation latent space. Our method allows us to perform downstream tasks like depth estimation and semantic segmentation by simply transferring linear heads trained on the RGB teacher. Most significantly, **REALM** enables the direct, zero-shot application of complex, frozen image-trained decoders, such as MASt3R, to raw event data. We demonstrate state-of-the-art performance in wide-baseline feature matching, significantly outperforming specialized architectures.
 
 <p align="right">(<a href="#readme-top">back to top</a>)</p>