Added UFM to research

Author: infinity1096

_posts/2025-06-12-ufm.md

@@ -0,0 +1,23 @@
+---
+layout: post
+title: "UFM: A Simple Path towards Unified Dense Correspondence with Flow"
+date: 2025-06-12 10:35:00
+categories: research
+description: "UFM is a simple, end-to-end trained transformer model that directly regresses pixel displacement images (flow) & covisibility which can be applied to both optical flow and wide-baseline matching tasks with high accuracy and efficiency."
+author: "Yuchen Zhang"
+published: true
+redirect: "https://uniflowmatch.github.io/"
+show_sidebar: false
+# slim_content_width: true
+permalink: /ufm_redirect/
+image: /img/posts/2025-06-12-ufm/ufm_teaser.gif
+datatable: true
+title_image: None
+hero_image: /img/posts/2025-06-12-ufm/ufm_teaser.gif
+hero_height: is-large
+remove_hero_title: false
+menubar_toc: false
+tags: Perception, Learning, SLAM
+---
+
+UFM is a simple, end-to-end trained transformer model that directly regresses pixel displacement images (flow) & covisibility which can be applied to both optical flow and wide-baseline matching tasks with high accuracy and efficiency.