diff --git a/assets/css/main.css b/assets/css/main.css
index bfd448e..eec6f7d 100644
--- a/assets/css/main.css
+++ b/assets/css/main.css
@@ -277,7 +277,7 @@ d-article d-contents nav ul { padding-left: 1em; margin-top: 0; margin-bottom: 6
 
 d-article d-contents nav ul li { margin-bottom: 0.25em; }
 
-d-article d-contents .figcaption { line-height: 1.4em; }
+d-article d-contents .figcaption { line-height: 1.4em; font-size: 0.8em;}
 
 d-article d-contents toc-line { border-right: 1px solid var(--global-divider-color); grid-column: toc-line; }
 
diff --git a/index.html b/index.html
index e5a24dd..769b7bd 100644
--- a/index.html
+++ b/index.html
@@ -174,6 +174,7 @@ <h5>NJU Vision Lab</h5>
 <hr>
 
 <h4 id="tldr">TLDR</h4>
+
 <p>This paper proposes a novel ρ-Vision to directly perform high-level semantic understanding and low-level compression using RAW images. The framework is demonstrated to provide better detection accuracy and compression than RGB-domain counterparts and is shown to be able to generalize across different camera sensors and task-specific models. Additionally, it has the potential to reduce ISP computation and processing time.</p>
 
 <center>
@@ -204,6 +205,8 @@ <h4 id="overview">Overview</h4>
 
 <h4 id="results">Results</h4>
 
+<h5 id="results" style="text-align: center;">RAW-Domain Detection</h5>
+
 <center>
     <figure class="col-md-8 col-md-offset-2">
     <img src="assets/img/det_zero_shot_results.jpg" class="img-responsive" style="max-width: 100%;" alt="zero-shot RAW detection results">
@@ -222,6 +225,8 @@ <h4 id="results">Results</h4>
     </figure>
 </center>
 
+<h5 id="results" style="text-align: center;">RAW-Domain Segmentation</h5>
+
 <center>
     <figure class="col-md-8 col-md-offset-2">
     <img src="assets/img/seg_zero_shot_results.jpg" class="img-responsive" style="max-width: 100%;" alt="zero-shot RAW segmentation results">