From 89a9f229c3e0e1038cc9274766c51fad6a930e38 Mon Sep 17 00:00:00 2001
From: JuliaKostiunina <i.kostiunina@uzum.com>
Date: Sun, 31 May 2026 12:59:42 +0200
Subject: [PATCH] Updated project card, added link to github

---
 index.html | 25 +++++++++++++------------
 1 file changed, 13 insertions(+), 12 deletions(-)

diff --git a/index.html b/index.html
index 7ceb546..7d59476 100644
--- a/index.html
+++ b/index.html
@@ -136,23 +136,24 @@ <h3>Probing V-JEPA 2: What Does a Video Model Actually See?</h3>
           </article>
 
           <article class="project-card">
-            <div class="teaser" role="img" aria-label="Two hands playing rock-paper-scissors, but one holds a banana instead of a valid sign, illustrating anomaly detection.">
+            <div class="teaser" role="img" aria-label="Visual product search for marketplace items using image embeddings, segmentation, and on-device retrieval.">
               <img src="assets/group_B.png" alt="" style="position:absolute; inset:0; width:100%; height:100%; object-fit:cover; z-index:2;">
               <span class="teaser-label" style="z-index:3;">Group B</span>
             </div>
             <div class="project-content">
-              <p class="project-meta">Vision-language embeddings, segmentation, on-device retrieval</p>
-              <h3>One Photo, Many Aisles: Visual Search Across a Heterogeneous Marketplace</h3>
+              <p class="project-meta">Visual search, marketplace retrieval, segmentation, Android demo</p>
+              <h3>Visual Search for Marketplace</h3>
               <p class="project-abstract">
-                A single foundation model rarely knows a sofa, a smartphone and a floral dress equally well — yet a real
-                marketplace catalog mixes all of them on the same shelf. This project studies where general-purpose visual
-                encoders like CLIP and SigLIP 2 stop being enough for e-commerce retrieval, and what has to be rebuilt
-                around them when the catalog is not one domain but twenty. Each query is first stripped of its context —
-                mannequins, human models, living-room scenes, studio gradients — then routed to a category-specific
-                expert whose embeddings are reranked with the fine-grained color and texture cues that generic models
-                quietly discard. The whole pipeline is then compressed into an on-device Android demo, raising a second
-                question the paper versions of Google Lens rarely address in the open: how much of a foundation-model
-                retrieval system actually survives when it has to run on a phone?
+                This project explores visual product search for marketplace applications, where users can search for visually similar
+                items using a single photo. I build a retrieval pipeline based on vision-language embeddings, object segmentation,
+                category-aware routing, and color-aware reranking to improve search quality for fashion and marketplace-style product
+                images. The final system is demonstrated in an Android app that performs on-device image-based retrieval over a local
+                product catalog.
+              </p>
+              <p>
+                <a href="https://github.com/siiena25/ImageSearch" target="_blank" rel="noopener noreferrer">
+                  GitHub / Code
+                </a>
               </p>
               <label class="project-toggle-label">
                 <input class="project-toggle" type="checkbox" aria-label="Toggle full project pitch">