BiaPy: accessible deep learning workflows YouTube video by I2K Conference

Did you miss our workshop on 💫 Halfway to I2K 2025 💫(www.i2kconference.org)? Worry no more! The video is finally out 👇

youtu.be/dFLtKHWlYdQ?...

Check it out and learn about BiaPy and its latest news 🤗 with @iarganda.bsky.social and @danifranco93.bsky.social 🤩

This is a new step for accessible, FAIR bioimage AI 🚀

Try the tool 👉 imaging.usegalaxy.eu?tool_id=tool...

Contribute your models 🧑‍🔬 And join the community making bioimage analysis open & reproducible!

#BiaPy #GalaxyProject #BioImageAnalysis #Bioimaging

a giraffe standing in front of a brick building that says headlife.camorange on the bottom ALT: a giraffe standing in front of a brick building that says headlife.camorange on the bottom

🤝 And there’s more! Thanks to this integration, you can also tap into the 🦒 BioImage Model Zoo 🧬.

Pretrained models can now be run straight through Galaxy with BiaPy, bringing community AI models to your data in just a few clicks.

#BioImageModelZoo #OpenScience

Screenshot of the Galaxy application in a browser, showing the BiaPy tool selected and its parameter selection options.

🛠️ We wrapped BiaPy as a Galaxy tool with a ✨ wizard-style interface (just like BiaPy’s GUI).

That means:
✔️ Step-by-step experiment setup
✔️ Launch training / inference jobs with BiaPy under the hood
✔️ Share & reproduce pipelines the Galaxy way 🌌

🚀 Big news: BiaPy meets @galaxyproject.bsky.social! ⛴️🤝🌌

You can now run BiaPy directly inside Galaxy, making deep learning for bioimage analysis easier, reproducible & accessible to everyone.

👉 imaging.usegalaxy.eu?tool_id=tool...

#BioImageAnalysis #GalaxyProject #AI4Life #Bioimaging

a man in a black and white sweater is standing in a living room and says `` seriously thanks so much '' . ALT: a man in a black and white sweater is standing in a living room and says `` seriously thanks so much '' .

So, one more time, 💫 thanks everybody 💫 for making this happen: from our developers 🛠️ to our users 👥.

Huge thanks and keep it up! 💪

In the meantime, we have more than 46 pretrained models from ⛴️ BiaPy available now in the 🦒 Bioimage Model Zoo 🫶 @ai4life.bsky.social

🔗 Have a look at: bioimage.io#/models

As usual, they are accesible from BiaPy's GUI 🖥️, Colab notebooks 📔 and our API 🤓

We fixed many issues ⚠️ and bugs 🐛 reported by our community 📩. You don't know how happy that makes us 😍

As an example, we can now reuse the predictions of a workflow and execute only the post-processing step with different parameters, which is much faster ⏩

biapy.readthedocs.io/en/latest/ge...

Development of Open-Source Deep Learning Tools for Bioimage Analysis

Special kudos to our fantastic summer intern Ibai Mayoral 👨‍🎓 from the @dipcehu.bsky.social who has made an incredible job at implementing all those methods now within BiaPy's core code 🤩

dipc.ehu.eus/en/career/in...

We have removed ✂️ our dependency to the unmaintained 'imgaug' library for data augmentation methods, which means we can now easily execute all our workflows in Google Colab without restarting 🔄 the session, yay! 🤗

bsky.app/profile/biap...

So no more this 👇

⛴️ BiaPy just reached version 3.6.1 💫 (latest release)!

github.com/BiaPyX/BiaPy...

➡️ v3.6.1 came right after (the more dense) v3.6.0 to fix some small bugs 🐞

Let me tell you a bit about what's new 👇

Yay! Well done @danifranco93.bsky.social 🤗 and long life to @ai4life.bsky.social 🤩

By democratizing bioimage analysis, BiaPy is paving the way for faster, more inclusive scientific breakthroughs, bringing us closer to understanding the building blocks of life.

mappingignorance.org/2025/05/15/biapy-analisi...

So happy to see #BiaPy in action! 🤗

Thanks a lot @jwpylvanainen.bsky.social 🤩

Right! That was one of our first options 🤗 However, since we need to extend many methods to 3D images, it seems more reasonable for us to write our own code and reduce the number of dependencies we already have 😇

Sure! We believe it is imgaug and its dependencies, which have not been updated in a while (5 years). So we're trying to get rid of it by writing our own code for the data augmentation methods we use from that library.

a man with dreadlocks and a beard is wearing a pirate hat and pointing at something . ALT: a man with dreadlocks and a beard is wearing a pirate hat and pointing at something .

Kudos to you, my captain! 🦸‍♂️

a shiba inu dog is smiling while sitting in a cage and looking at the camera . ALT: a shiba inu dog is smiling while sitting in a cage and looking at the camera .

Yay! 🥰

Thank you! 😊

a black cat is sitting in front of a laptop computer . ALT: a black cat is sitting in front of a laptop computer .

We're working on eliminating those dependencies 😅

The big deal is numpy. Some of the dependencies of BiaPy cannot work with numpy 2.*, so we have to remove the Colab's version of numpy and install numpy==1.26.*. And doing this, we're forced to restart the runtime for the changes to take effect 😬

BiaPy: accessible deep learning on bioimages - Nature Methods Nature Methods - BiaPy: accessible deep learning on bioimages

🙏 Thanks to
@naturemethods.bsky.social, @ritastrack.bsky.social, the reviewers, and the awesome bioimage analysis community.
We can't wait to see what you build with BiaPy! 🚀🧠
🔗 www.nature.com/articles/s41...
#BiaPy #BioimageAnalysis #DeepLearning

🔗 Want to try BiaPy?
Check out the full documentation and tutorials here:
👉 biapy.readthedocs.io
👉 www.youtube.com/@BiaPyX
Contribute, test, and join the community — BiaPy is open source! (6/7)
#OpenScience #Microscopy #AI

Screenshot of the BiaPy site showing the developing team: Daniel Franco-Barranco Jesús A Andrés-San Román Ivan Hidalgo-Cenalmor Lenka Backová Aitor González-Marfil Clément Caporal Anatole Chessel Pedro Gómez-Gálvez Luis M Escudero Donglai Wei Arrate Muñoz-Barrutia Ignacio Arganda-Carreras

🧬 BiaPy was built by an international collaboration between researchers from @upvehu.bsky.social, @dipcehu.bsky.social , @uc3m.es and others — led by
@iarganda.bsky.social and @arratemunoz.bsky.social.
A huge thanks to everyone who contributed! 🌟 (5/7)

Visual description of the multiple options available to interact with BiaPy: GUI (with friendly wizard), Google Colab and Jupyter notebooks, Docker container and terminal (library use).

Screenshot of the new BioImage.IO site, showing BiaPy as one of the available tools and community partners.

💡 How is BiaPy different?
• Full multi-GPU support for training and inference
• Efficient memory management for very large datasets (MBs → TBs)
• Single-container installation via Docker
• Integration with @BioImageModelZoo for easy model sharing and reuse!
#DeepLearning (4/7)

Overview of BiaPy's available workflows: semantic segmentation, single-image super-resolution, image denoising, self-supervised learning, instance segmentation, object detection, image classification and image-to-image translation.

⚙️ What can you do with BiaPy?
✅ Instance and semantic segmentation
✅ Object detection
✅ Image classification
✅ Image denoising & super-resolution
✅ Self-supervised learning
✅ Image-to-image translation
From 2D images to full 3D brain scans! 🧠 (3/7)

Animated gif showing an overview of BiaPy's features: its community inspiration, the diverse input data and analysis tasks, and its backend (PyTorch) and multiple frontends (GUI, notebooks, library).

🔍 What is BiaPy? 🚢
BiaPy is an open-source framework that simplifies the use of deep learning for microscopy 🔬 and biomedical images 🧪, offering intuitive interfaces, zero-code notebooks, and scalable workflows.
Built for scientists 🧑‍🔬 — no deep programming knowledge required! (2/7)
#AI #Bioimaging

BiaPy: accessible deep learning on bioimages - Nature Methods Nature Methods - BiaPy: accessible deep learning on bioimages

🚀 Excited to share that our work on BiaPy has been published today in Nature Methods!
BiaPy makes deep learning accessible for bioimage analysis, from small datasets to terabyte-scale 3D images. 👇 (1/7)
🔗 www.nature.com/articles/s41...
#DeepLearning #BioimageAnalysis #OpenSource #BiaPy

a cartoon opossum is standing on a dirt road and saying `` thank you '' . ALT: a cartoon opossum is standing on a dirt road and saying `` thank you '' .

We understand this is not ideal and we're working 🛠️ on updating BiaPy's ⛴️ dependencies so we can avoid restarting the notebooks in the near future. We'll keep you posted 📮, thanks for your patience!

Screenshot of a BiaPy notebook showing the "Play to reinstall Colab libraries so they are compatible with BiaPy" cell. The cell has been executed and Colab shows a message stating "Your session crashed for an unknown reason."

You need to run that cell and completely wait until it has finished, since it will restart 🔄 the notebook. You will see the message "Your session crashed for an unknown reason." Keep calm and carry on 🤓 You can execute the rest of the notebook as usual 👍