Revolutionizing Computer Vision: LightlyTrain's Self-Supervised Pretraining
LightlyTrain is transforming the landscape of computer vision by enabling developers to train high-performance models using unlabeled data. This innovative approach not only reduces labeling costs but...
Seeing @neauoire's paper computing explorations, I was reminded of an old idea I proposed at PaperCamp LDN back in 2009: Originally this was vaguely about using origami to create multi-purpose AR markers, but I then extended it to other use cases, some of which could be adapted and be even relevant today, e.g. as a form of proof-of-work, protection against AI crawlers or other form of access control.
Some possible approaches for different use cases:
1) Ask users to perform or series of simple folds and then check results by validating fold lines in the flat sheet paper via computer vision
2) Same as #1, but perform shape recognition/validation of fully folded result
3) Unfold a pre-shared (and pre-folded) origami object to check result by validating fold lines via computer vision
4) Provide instructions for multiple origami creations to create uniquely identifiable objects for computer vision based interactive environments
Number 1-3 are more or less about forms of proving physicality, work, membership. Number 4 is more about using origimi as fiducial markers for general interactions
More ideas/summary from that event:
https://adactio.com/journal/1546
cc/ @adactio
Unlocking colonial records with Artificial Intelligence. Achieving the automated transcription of large-scale 16th and 17th-century Latin American historical collections https://www.tandfonline.com/doi/abs/10.1080/20548923.2025.2484828 #AI #ComputerVision #archives
https://www.europesays.com/1988179/ Exploring Artificial Intelligence Applications in Agriculture #AI #ArtificialIntelligence #ComputerVision #ExpertSystems #MachineLearning #NaturalLanguageProcessing #PennStateExtension #ReinforcementLearning #Robotics
New article explores the work of Nestoras Papanikolopoulos, a pioneer of digital art in Greece since 1984. A vital step in recognizing early computer-based art and its marginalization in official discourse.
https://dahj.org/article/nestoras-papanikolopoulos
#DigitalArt #MediaArt #ComputerArt #ArtHistory #GreekArt #DigitalHeritage #DigitalHumanities #ComputerVision #DigitalArtHistory
Google’s New AI Mode Will Answer ‘Nuanced’ Questions About Your Photos https://petapixel.com/2025/04/08/googles-new-ai-mode-will-answer-nuanced-questions-about-your-photos/ #computervision #deeplearning #generativeai #googlegemini #googlesearch #multimodalai #Technology #googlelens #aimode #google #News
Today on OpenCV Live, we offer a reprieve from seeing your retirement fund slowly disappear in the form of an Introduction to Vision Language Models! It's a sneak peek of Satya's Embedded Vision Summit Workshop topic https://www.youtube.com/watch?v=trYjGml_ouk #OpenCV #ComputerVision #AI
Our #image #analysis #pipeline is ready to go into #pruduction.
It combines #AI and #forensic techniques to assess authenticity. It first extracts EXIF/XMP #metadata to detect real camera data forcing a “photograph” classification if present. Then, a CLIP based preprocessor (#computervision) classifies images into categories: photograph, meme, infographic, comic/cartoons, or text only.
Custom configurations (#development) adjust weights so that, for example, memes/text only images discount AI model influence while forensic tests (ELA, FFT, noise, edge sharpness, text detection) (#science, #math) quantify artifacts. #C2PA data is also considered. All evidence is combined to yield a final decision which will be the “confidence percentage” data that will be included in every post moving forward.
We just released wasmVision 0.3.1 as a quick followup to what was learned/fixed at #wasmio25 thanks everyone who came to the talk or spoke to @deadprogram afterwards!
Changelog here:
https://github.com/wasmvision/wasmvision/releases/tag/v0.3.1
Thanks to everyone who came to my talk at WASM I/O yesterday about @wasmvision
As promised, slides are now available here https://deadprogram.github.io/wasmio2025/#/
#wasmio25 #wasm #computervision #golang #tinygo #rust #clang #opencv
Love it when #rstats package docs have a sense of humour
"FAQ
Can I take selfies with my webcam using Rvision?
You certainly can, but ask yourself first whether you should…"
wasmVision 0.3.0 is out! We have some exiting new features for you such as MCP server support, and experimental GPU acceleration for vision models. Performance and stability improvements too. Go get it right now!
#wasm #computervision #opencv #golang #tinygo #rust #clang #mcp #cuda
https://github.com/wasmvision/wasmvision/releases/tag/v0.3.0
𝗨𝗻𝘀𝗲𝗿 𝗝𝗼𝘂𝗿𝗻𝗮𝗹𝗶𝘀𝘁-𝗶𝗻-𝗥𝗲𝘀𝗶𝗱𝗲𝗻𝗰𝗲-𝗣𝗿𝗼𝗴𝗿𝗮𝗺𝗺 𝗴𝗲𝗵𝘁 𝗶𝗻 𝗱𝗶𝗲 𝗻ä𝗰𝗵𝘀𝘁𝗲 𝗥𝘂𝗻𝗱𝗲
Das Programm bietet Journalist:innen die Möglichkeit, in einem drei- bis sechsmonatigen, bezahlten Aufenthalt in Tübingen zu einem selbst gewählten Thema zu recherchieren.
Von uns gibt es Einführungen in Themen wie #MachineLearning, #ComputerVision und #Robotik
Bitte bewerben Sie sich mit kurzem Ideenpapier, Anschreiben&Lebenslauf (DE o. ENG) 𝗯𝗶𝘀 𝟮𝟰.𝟬𝟰.𝟮𝟬𝟮𝟱
Bewerbungen bitte per E-Mail an janis.fischer@cyber-valley.de
OpenCV is hosting our first-ever official conference this May 12! Join us in San Jose for The OpenCV-SID Conference on Computer Vision & AI (OSCCA) for exclusive talks, community fellowship, and more. OpenCV CEO Dr. Satya Mallick will see you there! Register today: https://www.displayweek.org/event-enhancements/oscca/ #OpenCV #computervision #AI #OpenSource
We’re hosting another LIVE Q&A session for Bin-Picking Challenge participants with members of the BPC team. It begins Friday at 8am Pacific time. Join us!
Attend: https://us02web.zoom.us/webinar/register/WN_k8gaWuldQn6_oqeq_v9b0Q
The recording of the webinar will be available on our YouTube channel at http://youtube.com/@opencvofficial #OpenCV #AI #ComputerVision #OpenSource
Roboflow has launched RF-DETR, a real-time object detection model..
#AI #MachineLearning #ComputerVision #VisionAI #ObjectDetection #Roboflow #RFDETR #EdgeAI #DeepLearning #RealTimeAI #AIModels
Revolutionizing Mapping: Mozilla's AI Blueprint for OpenStreetMap
Mozilla.ai has unveiled an innovative AI Helper Blueprint that harnesses computer vision to enhance the OpenStreetMap community's mapping efforts. By integrating models like YOLOv11 and SAM2, this ini...
https://news.lavx.hu/article/revolutionizing-mapping-mozilla-s-ai-blueprint-for-openstreetmap
Want to see what you can do with computer vision and WebAssembly? Come to my talk next week at WASM I/O...
#wasmio25 #wasm #webassembly #computervision #golang #tinygo #rustlang #clang
https://2025.wasm.io/sessions/seeing-eye-to-eye-computer-vision-using-wasmvision/
Revolutionizing Stereo Depth Estimation with FoundationStereo: A Leap Toward Zero-Shot Generalization
The introduction of FoundationStereo marks a significant advancement in stereo depth estimation, overcoming challenges of zero-shot generalization with a robust foundation model. By leveraging a large...
