<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>Modality on Mohit Dulani</title>
    <link>https://complete-dope.github.io/codex/tags/modality/</link>
    <description>Recent content in Modality on Mohit Dulani</description>
    <generator>Hugo -- 0.146.0</generator>
    <language>en</language>
    <lastBuildDate>Fri, 06 Feb 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://complete-dope.github.io/codex/tags/modality/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>DL (image modality)</title>
      <link>https://complete-dope.github.io/codex/posts/dl-image_modality/</link>
      <pubDate>Fri, 06 Feb 2026 00:00:00 +0000</pubDate>
      <guid>https://complete-dope.github.io/codex/posts/dl-image_modality/</guid>
      <description>&lt;p&gt;Resolution means a lot for image domains an upscale from 512 to 1204 can make or break your AI model&lt;/p&gt;
&lt;p&gt;U2Net still works better for segmentation than most of the SAM3 models when it comes to high defined smooth edges.&lt;/p&gt;
&lt;p&gt;SAM models are data hungry, need a lot data to do something nice from it&lt;/p&gt;
&lt;p&gt;Convolution operation and kernels are underrated, a lot can be done if the filter / kernel values are set correctly&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
