sivaIITM commited on
Commit
5086618
·
verified ·
1 Parent(s): 4ea2a7b

Update index.html

Browse files
Files changed (1) hide show
  1. index.html +43 -4
index.html CHANGED
@@ -1,11 +1,50 @@
1
- <!DOCTYPE html>
2
  <html>
3
  <head>
4
  <meta charset="utf-8">
5
  <title>PySIFT: GPU-Resident Deterministic SIFT</title>
6
- <meta http-equiv="refresh" content="0; url=https://github.com/SivaIITM/PySIFT">
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
  </head>
8
  <body>
9
- <p>Redirecting to <a href="https://github.com/SivaIITM/PySIFT">GitHub repository</a>...</p>
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  </body>
11
- </html>
 
1
+ <!DOCTYPE html>
2
  <html>
3
  <head>
4
  <meta charset="utf-8">
5
  <title>PySIFT: GPU-Resident Deterministic SIFT</title>
6
+ <style>
7
+ body { font-family: system-ui, sans-serif; max-width: 800px; margin: 40px auto; padding: 0 20px; color: #333;
8
+ }
9
+ h1 { color: #1a1a1a; }
10
+ .badge { display: inline-block; margin-right: 8px; }
11
+ .links { margin: 24px 0; }
12
+ .links a { display: inline-block; padding: 10px 20px; margin: 4px 8px 4px 0; background: #2ecc71; color:
13
+ white; text-decoration: none; border-radius: 6px; font-weight: bold; }
14
+ .links a.paper { background: #b31b1b; }
15
+ .links a.pypi { background: #3775a9; }
16
+ table { border-collapse: collapse; margin: 20px 0; }
17
+ th, td { border: 1px solid #ddd; padding: 8px 14px; text-align: left; }
18
+ th { background: #f5f5f5; }
19
+ code { background: #f0f0f0; padding: 2px 6px; border-radius: 3px; }
20
+ </style>
21
  </head>
22
  <body>
23
+ <h1>PySIFT</h1>
24
+ <p><strong>GPU-Resident Deterministic SIFT for Deep Learning Vision Pipelines</strong></p>
25
+
26
+ <div class="links">
27
+ <a class="paper" href="https://arxiv.org/abs/2605.17869">arXiv Paper</a>
28
+ <a href="https://github.com/SivaIITM/PySIFT">GitHub Code</a>
29
+ <a class="pypi" href="https://pypi.org/project/staysift/">pip install staysift</a>
30
+ </div>
31
+
32
+ <p>A pure-Python, GPU-resident SIFT implementation that matches OpenCV SIFT accuracy while running <strong>26%
33
+ faster end-to-end</strong> with <strong>4x matching speedup</strong>. Zero-copy DLPack interop keeps tensors on the
34
+ GPU across the full pipeline.</p>
35
+
36
+ <table>
37
+ <tr><th>Benchmark</th><th>Metric</th><th>PySIFT vs OpenCV</th></tr>
38
+ <tr><td>HPatches</td><td>MMA@10</td><td>+2.2pp</td></tr>
39
+ <tr><td>IMC Phototourism</td><td>Inliers/pair</td><td>303 vs 205 (+47%)</td></tr>
40
+ <tr><td>MegaDepth-1500</td><td>AUC@10</td><td>+5.6pp</td></tr>
41
+ <tr><td>ROxford5K</td><td>mAP</td><td>+7.5pp</td></tr>
42
+ </table>
43
+
44
+ <h3>Quick Start</h3>
45
+ <pre><code>pip install staysift
46
+ from pysift import PySIFT
47
+ sift = PySIFT()
48
+ keypoints, descriptors = sift.detectAndCompute(gray_image)</code></pre>
49
  </body>
50
+ </html>