Rainer SimonAI, LLM, IMMARKUS
sigmoid.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A social space for people researching, working with, or just interested in AI!
Administered by:
Server stats:
586active users
sigmoid.social: About · Profiles directory · Privacy policy
Mastodon: About · Get the app · Keyboard shortcuts · View source code · v4.4.3
#immarkus
1 post · 1 participant · 0 posts today
Rainer Simon<p>We are extending the multi-select options in <a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a>! Merge and subtract selections, batch-tag and batch-delete, and (new!) compare annotation data side-by-side. The multi-select panel is getting a small facelift, too!</p>
Rainer Simon<p>Quick update: Plugging your own transcription service into <a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a> via Hugging Face Spaces is now available for testing on our dev server!</p><p>It’s a bit DIY/hacky... But if you’re building your own HF Spaces, you’re probably ready for that 🙂 Ping me if you’d like to try it out!</p>
Rainer Simon<p>Back to work after a short break, and working on two new <a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a> extensions:</p><p>• Use HuggingFace Inference Providers as a transcription service<br>• For the adventurous: plug in your own HuggingFace Space as a transcription service!</p><p>The second one’s experimental... if you're running your own HF Space and want to try it with IMMARKUS, give me a shout! I'd love to collaborate and test it out.</p>
Rainer Simon<p>Because sometimes you want that extra bit of precision: Bezier curve drawing is coming—pixel-perfect annotations for <a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a>, <a href="https://vis.social/tags/liiive" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>liiive</span></a>, and all tools powered by <a href="https://vis.social/tags/Annotorious" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Annotorious</span></a></p>
Rainer Simon<p>It has come to my attention that some people (scandalously!) want to work with their image annotations OUTSIDE of <a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a>. If you're so inclined, you might enjoy our upcoming copy-and-paste feature!</p>
Rainer Simon<p>If you haven’t checked in on <a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a> lately (understandable—there’s been a lot going on!)—we’ve added even more transcription service options.</p><p>You can now run OCR or full-text transcription with a single click using:</p><p>• Anthropic Claude<br>• Azure Computer Vision<br>• Google Gemini<br>• Google Vision OCR<br>• LLaMA & Qwen via kluster.ai<br>• OCR.space<br>• OpenAI GPT<br>• Volcano Engine Doubao 1.5 Vision Pro</p><p>Try it out here: <a href="https://immarkus.xmarkus.org" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">immarkus.xmarkus.org</span><span class="invisible"></span></a></p><p><a href="https://vis.social/tags/OCR" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OCR</span></a> <a href="https://vis.social/tags/DigitalHumanities" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DigitalHumanities</span></a> <a href="https://vis.social/tags/ImageAnnotation" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ImageAnnotation</span></a> <a href="https://vis.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://vis.social/tags/IIIF" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IIIF</span></a></p>
Rainer Simon<p>Small enhancement to <a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a>’ Google Vision integration: You can now choose the level of detail when importing OCR results—words, paragraphs, or full blocks. </p><p><a href="https://vis.social/tags/OCR" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OCR</span></a> <a href="https://vis.social/tags/IIIF" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IIIF</span></a> <a href="https://vis.social/tags/ImageAnnotation" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ImageAnnotation</span></a> <a href="https://vis.social/tags/DigitalHumanities" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DigitalHumanities</span></a></p>
Rainer Simon<p>In our quest to make <a href="https://vis.social/tags/OCR" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OCR</span></a> as easy to use as possible, we've added two new services to <a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a>: Google Gemini and OpenAI GPT. (You’ll need to bring your own API keys to use them.)</p><p>Both return full-text transcriptions–no bounding boxes–so you’ll need to select a region before running them. Works with local images and any <a href="https://vis.social/tags/IIIF" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IIIF</span></a> source!</p>
Rainer Simon<p>We’re adding more OCR and image annotation services to <a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a>!</p><p>Now available: Google Vision API — which actually works really well with historical materials, too. 👀 (Note: you’ll need to bring your own API key.)</p><p><a href="https://vis.social/tags/DigitalHumanities" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DigitalHumanities</span></a> <a href="https://vis.social/tags/IIIF" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IIIF</span></a> <a href="https://vis.social/tags/OCR" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OCR</span></a> <a href="https://vis.social/tags/ImageAnnotation" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ImageAnnotation</span></a></p>
Rainer Simon<p>We’re working on an extension that lets you send images to OCR services right from your <a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a> workspace! </p><p>• Submit full images or selected regions<br>• Run multiple passes, preview results<br>• Import word- or line-level annotations<br>• Works with local files and <a href="https://vis.social/tags/IIIF" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IIIF</span></a></p><p>Got an OCR engine or map-text service we should support? Reach out!</p><p>Or just need a flexible tool to annotate historical images? Try IMMARKUS: <a href="https://immarkus.xmarkus.org" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">immarkus.xmarkus.org</span><span class="invisible"></span></a></p><p><a href="https://vis.social/tags/DigitalHumanities" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DigitalHumanities</span></a> <a href="https://vis.social/tags/IIIF" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IIIF</span></a> <a href="https://vis.social/tags/OCR" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OCR</span></a></p>
Rainer Simon<p>Aaaand… we’re already experimenting with bringing MapReader outputs into our <a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a> annotation tool—for validation, cleanup, correction, and further analysis. (Seen here: a map with ~2,500 transcribed labels.) <a href="https://vis.social/tags/ImageAnnotation" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ImageAnnotation</span></a> <a href="https://vis.social/tags/IIIF" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IIIF</span></a> <a href="https://vis.social/tags/Annotorious" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Annotorious</span></a></p>
Rainer Simon<p>Haven't explored IMMARKUS yet? Now’s a great time!</p><p>• New docs w/ clearer guides and (animated) screenshots<br>• Multi-image annotation workbench (IIIF or local files)<br>• Use AI tools for fast region selection<br>• Build + explore your own ontology & knowledge graph<br>• 100% local, open source, in your browser</p><p><a href="https://immarkus.xmarkus.org" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">immarkus.xmarkus.org</span><span class="invisible"></span></a></p><p><a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a> <a href="https://vis.social/tags/ImageAnnotation" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ImageAnnotation</span></a> <a href="https://vis.social/tags/IIIF" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IIIF</span></a> <a href="https://vis.social/tags/DigitalHumanities" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DigitalHumanities</span></a> <a href="https://vis.social/tags/OpenSource" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OpenSource</span></a></p>
Rainer Simon<p><a href="https://vis.social/tags/IMMARKUS" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IMMARKUS</span></a> update: our next release will complete the toolbox—bringing the full set of <a href="https://vis.social/tags/computervision" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>computervision</span></a> and <a href="https://vis.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a>-powered drawing tools!</p><p>• Smart Scissors – trace edges, Photoshop-style.<br>• Edge Snap – snap to corners & lines for clean cutouts.<br>• Auto Select – smart one-click object selection.</p><p>Want to know more? Ping us! <a href="https://vis.social/tags/IIIF" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>IIIF</span></a> <a href="https://vis.social/tags/ImageAnnotation" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ImageAnnotation</span></a> <a href="https://vis.social/tags/DigitalHumanities" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>DigitalHumanities</span></a></p>
Continued thread
And another one! 
A smart selection tool powered by the "Intelligent Scissors" algorithm from OpenCV.js. This should make complex polygon annotation a breeze! Coming to #IMMARKUS (very) soon. #ImageAnnotation #IIIF
My colleagues from the infrastructurelives.eu project have updated the documentation for #IMMARKUS and it's great! Up to date with all the latest features. Lots of detailed instructions, screenshots, and generally much improved over my initial braindump from a year ago.
Working on some new drawing features 
coming soon to #IMMARKUS: Boolean operations (merge and subtract shapes)! Plus: shapes can now have holes. #JavaScript #Annotorious #IIIF #OpenSeadragon
Wait... is that a new button in #IMMARKUS? 
Yes! Click the magic wand for automatic object selection, powered by #SegmentAnything. Just point, click, and let AI do the work! Try it out and let us know what you think. #DigitalHumanities #IIIF
Update: Smart object selection in zoomable images with #OpenSeadragon and #SegmentAnything! The next key challenge: Creating a seamless workflow—picking objects, refining shapes, and manually enhancing the outline. Stay tuned! #IMMARKUS #IIIF #DigitalHumanities #Maps
TrendingLive feeds
Mastodon is the best way to keep up with what's happening.
Follow anyone across the fediverse and see it all in chronological order. No algorithms, ads, or clickbait in sight.
Create accountLoginDrag & drop to upload