<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Polar Transformer Networks · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/1709.01889/</link><description>Collected HN comments and submissions linking this paper.</description><language>en</language><lastBuildDate>Mon, 02 Nov 2026 11:16:22 GMT</lastBuildDate><atom:link href="https://hackerpaper.kaipfister.de/p/1709.01889/feed.xml" rel="self" type="application/rss+xml" /><item><title>Polar Transformer Networks — mentioned by magicalhippo</title><link>https://news.ycombinator.com/item?id=17284886</link><guid isPermaLink="false">urn:hackerpapers:mention:1709.01889:17284886</guid><pubDate>Mon, 11 Jun 2018 15:09:37 GMT</pubDate><description>&lt;p&gt;Having dabbled with image processing, your comment reminded me of the Fourier Mellin transform, which can be used for translation, rotation and scale invariant feature detection. I did a quick search and came up with this paper[1], and while they don&amp;#x27;t use the Fourier-Mellin transform, they do use a log-polar transform[2], where rotation and scaling are transformed to translations. This, they claim, result in much improved classification of rotated and scaled data. I haven&amp;#x27;t been keeping up on the latest in the ML field tho, so maybe it&amp;#x27;s rubbish, but like you I&amp;#x27;d be surprised if some extra…&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/1709.01889/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.CV</category></item></channel></rss>
