Don't bother parsing: Just use images for RAG

https://www.morphik.ai/blog/stop-parsing-docs

Behold, the majestic #PlutoFilter, the C library that promises to filter your images with zero allocations—because who needs memory, right? Meanwhile, GitHub is busy dreaming up ways for #AI to write your code, because clearly, humans have already reached peak #coding perfection.
https://github.com/sammycage/plutofilter #imageprocessing #memorymanagement #HackerNews #ngated

I probably have discovered a thousand bugs in softwares developed by people other than myself, but this time, for the first time in my life, I found a bug without producing a buggy phenomenon first but by staring at the source codes and deducing mathematical formulas by hand. That was an amazing experience. I wrote a blog article about it: https://ulysseszh.github.io/programming/2025/07/17/blur-filter.html

Deep adaptive learning predicts and diagnoses CSVD-related cognitive decline using radiomics from T2-FLAIR: a multi-centre study

Patient enrolment and baseline characteristics A total of 783 su…
#NewsBeep #News #US #USA #UnitedStates #UnitedStatesOfAmerica #Artificialintelligence #AI #ArtificialIntelligence #Biomedicine #biotechnology #Cognitiveageing #Cognitiveneuroscience #Computationalneuroscience #general #Imageprocessing #Medicine/PublicHealth #Technology
https://www.newsbeep.com/us/12439/

Lossless Float Image Compression

https://aras-p.info/blog/2025/07/08/Lossless-Float-Image-Compression/

How that I have your attention I would like to exploit it for some work stuff. #RemoteSensing #ImageProcessing communities, I have a real quest. We're segmenting shallow underwarter structures in these archival aerial photos (grayscale, single-band) and the segmentation algorithm picks up... the waves on the water surface, not reefs and such. This gives an artefactual vergence (direction) that is not present in the layer we are interested in, i.e. sea floor. This is SVM in eCognition. Don't mention segmentation in ArcGIS (unless you're a powerful wizard, in that case I bow).
We also have a custom ML model but it's not that good either (yet?). How to get rid of them waves?! #geoscience #EarthObservation

Louis-Jacques-Mandé Daguerre, who died OTD in 1851, was an early photographer who invented the daguerreotype process https://cromwell-intl.com/3d/histogram/?s=mb #history #ImageProcessing #photography

I'm not sure if this process has a name : iteratively sharpen, then blur an image. Sounds counter-intuitive, but it gives you these results. (I guess it's Gray-Scott-esque??)
- may need to balance the 2 steps - here i used 2x Pillow's "ImageFilter.SHARPEN" + 1x "ImageFilter.GaussianBlur(radius=1)"
You can add noise each step (I didn't here).
- I used Python but you can set up actions in PS or Gimp
- need many iterations - these were 400.

Anyone know a proper name for this?
#imageprocessing

High Performance Image Sensor Processing Using FPGA [pdf]

https://oda.uni-obuda.hu/bitstream/handle/20.500.14044/10350/Gabor_S_Becker_ertekezes.pdf

FREE 50% storage space! Open-source bulk image compressor

As someone processing 1000+ images daily, I built #imageVision to solve compression headaches:
Drag-n-drop batch processing
Smart lossless compression (quality preserved!)
Up to 50% space savings

Get it: GitHub "imageVision"

via @dotnet : Local AI + .NET = AltText Magic in One C# Script

https://ift.tt/iFfJ2Gm
#AI #DotNet #AltText #CSharp #Accessibility #LocalModels #Ollama #ImageProcessing #Automation #GenerativeAI #MachineLearning #DevTools #TechForGood #SEO #UX #Coding #Programmin…

I asked ChatGPT, Claude, Grok, and deepseek to flip a png that contained text and then flip the text to be readable again. Deepseek can't handle images, Claude did sth crazy & useless output, Grok similarly, Chat just flipped the image but not the text.
Can this be done? How?

A rainbow around the full moon? I think it's called an ice bow or halo.

Actually, I see at least three features around last night's full moon: (1st photo)

1. (part of) a circular halo around the moon...which is colored red on the inside, blue on the outside

2. a moondog to the left of the moon, also on the circular halo

3. part of a colored halo (top center of frame) that appears to encircle the zenith. This is quite faint.

Don't get distracted by the internal camera lens reflections that are a bit above and right of center.

I processed the same photo in a different way to make stars more visible. The full moon is in the heart of Scorpio. The right portion of the frame contains Corvus the crow, and above that is Virgo.

This image of the moon, sky, stars...in a loose sense...is data. As I process the data differently, I can make different features easier to see.

Convert Photos to Atkinson Dithering

https://gazs.github.io/canvas-atkinson-dither/

Probando mi sistema de segmentación de imagen

Workflow CG 8 and 9 https://photo.m-j-s.net/blog/2025/06/workflow-cg-8-and-9/