<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Bonggi Cho</title>
    <description>The latest articles on DEV Community by Bonggi Cho (@chobonggi3).</description>
    <link>https://dev.to/chobonggi3</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4256457%2F9c5fd733-fd44-4c2b-a030-cba212b5fdef.png</url>
      <title>DEV Community: Bonggi Cho</title>
      <link>https://dev.to/chobonggi3</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/chobonggi3"/>
    <language>en</language>
    <item>
      <title>Traveling in Korea: reading menus, signs and notices with a phone camera</title>
      <dc:creator>Bonggi Cho</dc:creator>
      <pubDate>Sat, 21 Nov 2026 02:07:40 +0000</pubDate>
      <link>https://dev.to/chobonggi3/traveling-in-korea-reading-menus-signs-and-notices-with-a-phone-camera-1cgc</link>
      <guid>https://dev.to/chobonggi3/traveling-in-korea-reading-menus-signs-and-notices-with-a-phone-camera-1cgc</guid>
      <description>&lt;p&gt;Korea is easy to travel in until the moment everything is written in Korean. A menu on a wall, a notice taped to a shop door, a form on a counter, the label on a medicine packet you were handed at a pharmacy: you can point at things, but you cannot read them.&lt;/p&gt;

&lt;p&gt;I am a solo Android developer in Seoul, and I built a small free app for that. But most of this post is about the problem and the trades involved in reading text with a phone camera, because that part applies whatever app you use.&lt;/p&gt;

&lt;h2&gt;
  
  
  The three places Korean-only text catches a visitor
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Food. Menus often have pictures, but the prices, the portion size, the "spicy" warning and whether a dish contains pork or beef are in text.&lt;/li&gt;
&lt;li&gt;Signs and notices. "Closed today", "cash only", "elevator out of order", opening hours printed in a font size nobody over forty can read at arm's length.&lt;/li&gt;
&lt;li&gt;Paper handed to you. A receipt, a delivery slip, a prescription envelope, a form at a clinic. This is the case where translation is not a convenience but the difference between understanding and guessing.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  What to look for in a camera reader
&lt;/h2&gt;

&lt;p&gt;Four things decide whether a camera OCR app is useful on a trip.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Does it work without a connection? Roaming data is expensive, and subway platforms are exactly where you need to read something.&lt;/li&gt;
&lt;li&gt;Where does the photo go? If recognition happens on a server, the photo of your prescription is now on someone else's computer.&lt;/li&gt;
&lt;li&gt;How many taps from pocket to answer? A reader that needs an account, a login and a language picker before it shows anything is not going to be used at a restaurant table.&lt;/li&gt;
&lt;li&gt;Does it speak? A spoken translation is easier to follow while you are standing in front of a person who is waiting for your answer.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why on-device processing changes the experience
&lt;/h2&gt;

&lt;p&gt;The app I built, Photo Read, runs recognition and translation on the phone. The photo is not uploaded, and after each language file has been downloaded once, translation keeps working offline. That decision has a cost: the models ship with or get downloaded to the app, so the first use of a language needs an internet connection, and on-device models are not the largest ones available.&lt;/p&gt;

&lt;p&gt;What you gain is speed, privacy and the ability to work in a basement restaurant with no signal. On-device inference is also the reason the app can be free with ads and still keep a session short: no round trip to a server, nothing to bill per request.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the app does, concretely
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Take a photo of text: a menu, a sign, a document.&lt;/li&gt;
&lt;li&gt;The app recognises the text and translates it into Korean, English, Chinese or Japanese.&lt;/li&gt;
&lt;li&gt;One tap reads the translation out loud, after a short ad.&lt;/li&gt;
&lt;li&gt;One screen: four language buttons and one big capture button, no settings to get lost in.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Google Play: &lt;a href="https://play.google.com/store/apps/details?id=io.cho.photo_read" rel="noopener noreferrer"&gt;https://play.google.com/store/apps/details?id=io.cho.photo_read&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;It is free, ad-supported, has no account and no in-app purchase. The same person also publishes BboshaCam, a free camera app, but that is a different story.&lt;/p&gt;

&lt;h2&gt;
  
  
  Practical tips for reading Korean in the wild
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Fill the frame with the text block, not the whole menu board. Recognition works on what is in focus.&lt;/li&gt;
&lt;li&gt;Keep the phone parallel to the paper. Perspective is harder than blur.&lt;/li&gt;
&lt;li&gt;Photograph one paragraph or one column at a time. A page of dense Korean is a harder problem than three lines of it.&lt;/li&gt;
&lt;li&gt;Take the photo before you ask the question. Standing at a counter with a screen is easier for everyone than pointing at the wall twice.&lt;/li&gt;
&lt;li&gt;Treat the output as a strong hint, not a legal translation. Numbers and short phrases are reliable; a long complicated sentence is where machine translation stops being a translator and becomes a paraphrase.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  A note for other developers
&lt;/h2&gt;

&lt;p&gt;If you are building something similar, the constraint that improved the app most was deleting options. The first version had a settings screen and a language picker; every person I handed the phone to opened it, changed a language and got lost. Four fixed buttons and one capture button did more for real usage than any accuracy improvement.&lt;/p&gt;

&lt;p&gt;The second was treating offline as a feature rather than a fallback. The moment the app works with no signal, it stops being a tool for a planned task and becomes something you can reach for in the middle of a day.&lt;/p&gt;

&lt;p&gt;If you travel here and try it, I would like to know what it got wrong first. That is the most useful feedback I can get.&lt;/p&gt;

</description>
      <category>koreatravel</category>
      <category>android</category>
    </item>
    <item>
      <title>Two free little Android apps I built alone: on-device photo translation and a camera that fixes photos after you shoot</title>
      <dc:creator>Bonggi Cho</dc:creator>
      <pubDate>Fri, 20 Nov 2026 23:22:27 +0000</pubDate>
      <link>https://dev.to/chobonggi3/two-free-little-android-apps-i-built-alone-on-device-photo-translation-and-a-camera-that-fixes-22bi</link>
      <guid>https://dev.to/chobonggi3/two-free-little-android-apps-i-built-alone-on-device-photo-translation-and-a-camera-that-fixes-22bi</guid>
      <description>&lt;p&gt;I am a solo developer in Seoul. Outside my day job I ship small free Android apps, and two of them are now on Google Play. They have something in common: both exist because a built-in feature was one step short of what I needed.&lt;/p&gt;

&lt;p&gt;Below is what each one does, the technical choices behind it, and what I would do differently. Feedback is very welcome.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Photo Read — photo OCR plus translation, processed on the device
&lt;/h2&gt;

&lt;p&gt;Photo Read started with a simple annoyance: holding a prescription envelope or a delivery notice at arm's length and squinting. A magnifier app shows you the letters, but it cannot tell you what they mean, and most OCR apps upload the photo to a server and hand back text.&lt;/p&gt;

&lt;p&gt;What it does:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Take a photo of the text you want to read (menu, prescription envelope, sign, document).&lt;/li&gt;
&lt;li&gt;The app recognises the text and translates it into Korean, English, Chinese or Japanese.&lt;/li&gt;
&lt;li&gt;After one ad, it reads the translation out loud.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Implementation notes:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Recognition and translation run &lt;strong&gt;on the device&lt;/strong&gt;. The photo is not uploaded anywhere, which is also why the app keeps working when there is no connection: after the language pack is downloaded once, the app can translate a photo with the phone offline.&lt;/li&gt;
&lt;li&gt;One screen, four language buttons, one big capture button. That constraint was deliberate: the app is meant to stay usable for people who are not power users.&lt;/li&gt;
&lt;li&gt;Ads are the only monetisation. There is no in-app purchase and no account.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Google Play: &lt;a href="https://play.google.com/store/apps/details?id=io.cho.photo_read" rel="noopener noreferrer"&gt;https://play.google.com/store/apps/details?id=io.cho.photo_read&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;What I would do differently: the first version had a settings screen, and it got in the way of the one thing the app is for. Removing it was the single biggest improvement in usability.&lt;/p&gt;

&lt;h2&gt;
  
  
  2. BboshaCam — a camera app that treats editing as part of shooting
&lt;/h2&gt;

&lt;p&gt;BboshaCam is the opposite of the first app: no text, no translation, just photos. It exists because editing a photo in three different apps was too much friction for everyday pictures.&lt;/p&gt;

&lt;p&gt;What it does:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Four shooting modes: portrait, photo, landscape, night.&lt;/li&gt;
&lt;li&gt;After the shot, adjust brightness, colour tone, skin, sharpness and filters on the same screen.&lt;/li&gt;
&lt;li&gt;Free, ad-supported, Android 7.0 and up.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Implementation notes:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;The editor is part of the shooting flow rather than a separate gallery app. One screen: shoot, then tune the sliders that matter for people photos.&lt;/li&gt;
&lt;li&gt;Night mode and the sliders are tuned for indoor and evening phone photos, which is where a default camera usually disappoints.&lt;/li&gt;
&lt;li&gt;No account and no upload; the only third-party SDK is the ad library.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Google Play: &lt;a href="https://play.google.com/store/apps/details?id=io.cho.bbocam" rel="noopener noreferrer"&gt;https://play.google.com/store/apps/details?id=io.cho.bbocam&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  What I learned shipping two apps alone
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Pick a user you can picture.&lt;/strong&gt; "Someone squinting at a prescription envelope" is a better specification than "users who want OCR".&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Offline is a feature people feel.&lt;/strong&gt; With on-device processing, the app is fast, works abroad, and does not need a privacy policy paragraph about uploads.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;One screen beats a settings menu.&lt;/strong&gt; Both apps got better when I deleted options.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Free with ads is a real constraint.&lt;/strong&gt; Every extra network call and every heavy library shortens the session for someone on a cheap phone. On-device inference was not only a privacy decision.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;If you try either app, I would like to know what broke first. Comments here reach me, and the Play listing has the support address.&lt;/p&gt;

&lt;p&gt;(Posted by the developer. Both apps are free with ads; there is nothing to buy.)&lt;/p&gt;

</description>
      <category>android</category>
      <category>mobile</category>
      <category>showdev</category>
      <category>productivity</category>
    </item>
  </channel>
</rss>
