Embed Notice

HTML Code

<blockquote style="position: relative; padding-left: 55px;"><section><a href="https://social.alt-text.org/users/hannah/statuses/112672841435489331">Hannah Kolbeck 🏳️‍⚧️ (hannah@social.alt-text.org)'s status on Tuesday, 25-Jun-2024 03:10:31 JST</a><a href="https://social.alt-text.org/@hannah" title="hannah@social.alt-text.org"><img src="https://gnusocial.jp/avatar/147351-48-20230724062022.webp" width="48" height="48" alt="Hannah Kolbeck 🏳️‍⚧️" style="position: absolute; left: 0; top: 0;">Hannah Kolbeck 🏳️‍⚧️</a></section><article><p>I've been working on this for a while and improved some things yesterday. <a href="https://2d-ocr.glitch.me" rel="nofollow noreferrer">https://2d-ocr.glitch.me</a> takes an image with text in chunks and extracts them into an HTML table. More work needed on accessibility.</p><p>Still some issues such as with text grouping on more complex images that have possible solutions but probably beyond my focus. Also an issue is the use of Google's OCR API which was the only tool I could find a few years ago making this possible, local/free libraries might now be available.</p></article><footer><a rel="bookmark" href="https://gnusocial.jp/conversation/3286131#notice-6477989">In conversation</a><time datetime="2024-06-25T03:10:31+09:00" title="Tuesday, 25-Jun-2024 03:10:31 JST">about 5 months ago</time> <span>from <span><a href="https://social.alt-text.org/@hannah/112672841435489331" rel="external" title="Sent from social.alt-text.org via ActivityPub">social.alt-text.org</a></span></span><a href="https://social.alt-text.org/@hannah/112672841435489331">permalink</a><h4>Attachments</h4><ol><li><label><a rel="external" href="https://gnusocial.jp/attachment/2825528">A four panel comic. A bunch of cartoon people coded as chess pieces in simple black armor say to their king "Your majesty, the enemy has surrounded your position, we fear that there may be no escape." The king looks and sees a white horse, then a white bishop and castle, then says "Yep, let's surrender"</a></label><br><a href="https://cdn.masto.host/socialalttextorg/media_attachments/files/112/672/793/261/116/059/original/2b81fe111f8f45ba.png" rel="external">https://cdn.masto.host/socialalttextorg/media_attachments/files/112/672/793/261/116/059/original/2b81fe111f8f45ba.png</a></li><li><label><a rel="external" href="https://gnusocial.jp/attachment/2825529">The meme with text chunks in red boxes, as well as the cross on the kings hat and watermarks</a></label><br><a href="https://cdn.masto.host/socialalttextorg/media_attachments/files/112/672/796/396/364/861/original/b3140dd1372f1979.png" rel="external">https://cdn.masto.host/socialalttextorg/media_attachments/files/112/672/796/396/364/861/original/b3140dd1372f1979.png</a></li><li><label><a rel="external" href="https://gnusocial.jp/attachment/2825530">An HTML table with each cell holding the contents of one of the boxes, where the king's cross is translated as a "+" and the cells are slightly out of place relative to the parts of the meme. In the comic text is all caps, and it is in the box as well, probably need to optionally downcase</a></label><br><a href="https://cdn.masto.host/socialalttextorg/media_attachments/files/112/672/799/513/548/764/original/bea9c7a21db43a0f.png" rel="external">https://cdn.masto.host/socialalttextorg/media_attachments/files/112/672/799/513/548/764/original/bea9c7a21db43a0f.png</a></li><li><article><header><div>No result found on File_thumbnail lookup.</div><h5><a href="https://2d-ocr.glitch.me/">2D OCR</a></h5><div></div></header><div></div><footer></footer></article></li></ol></footer></blockquote>

Corresponding Notice

Embed this notice
Hannah Kolbeck 🏳️‍⚧️ (hannah@social.alt-text.org)'s status on Tuesday, 25-Jun-2024 03:10:31 JST Hannah Kolbeck 🏳️‍⚧️
I've been working on this for a while and improved some things yesterday. https://2d-ocr.glitch.me takes an image with text in chunks and extracts them into an HTML table. More work needed on accessibility.
Still some issues such as with text grouping on more complex images that have possible solutions but probably beyond my focus. Also an issue is the use of Google's OCR API which was the only tool I could find a few years ago making this possible, local/free libraries might now be available.
In conversationabout 5 months ago from social.alt-text.orgpermalink
Attachments

Public

Embed Notice

HTML Code

Corresponding Notice