Computer Vision news and analysis on Web Pulse: 3357 curated articles tracking the latest Computer Vision developments, tools, and research, updated continuously from vetted sources.
Generative AI for spatial content has rapidly evolved from basic 3D asset creation to interactive world models and AI-native CAD between May 2025 and June 2026, driven by releases from Meta, Google DeepMind, Tencent, and…
GPT Image 2 API defeated Seedream 5.0 Pro Image Editing 72.0 to 65.2 in an eight-task image editing test, winning four tasks outright with zero losses, demonstrating superior adherence to structure-heavy prompts involvin…
New advancements in Vision Language Action (VLA) models are achieving 39 times faster inference speeds than OpenVLA and 46 Hz throughput on edge platforms, reducing latency and costs through fewer action tokens and a vot…
Compass, a new AI methodology for prostate cancer detection, analyzes 2D micro-ultrasound studies using a multi-view approach that mimics clinical practice by integrating rotational video sweeps, outperforming traditiona…
Roboflow launched a product recognition AI system that combines object detection with text recognition to automate checkout validation and order packing, helping businesses eliminate human billing errors by verifying phy…
Researchers analyzed 2.6 billion sketches from 236 countries to study how visual representations of concepts vary across cultures, finding that visual similarities align 45% more with cultural distances than text-based m…
Henry Schein One built Image Verify, an AI-powered quality verification system on Amazon SageMaker AI that evaluates dental X-ray quality in real time at the point of capture. The system, which processes over 1.5 million…
LoopVera, an open-source runtime evidence ledger for debugging vision pipelines and agent workflows, is being developed to provide a see-tweak-diff-sign-off loop without replacing existing operators. The project is in ea…
Mistral released Robostral Navigate, an 8B-parameter AI model that uses a single RGB camera and natural-language instructions to navigate robots through indoor spaces. The model achieved a 76.6% success rate on the R2R-C…
Skipper has added HD camera quality profile switching, allowing the system to autonomously determine optimal camera settings. The update also centralizes runtime config defaults for improved organization and introduces A…
Researchers found that the Gemma 4 2B vision-language model achieved 93.65% accuracy in detecting fast radio bursts, nearly matching the specialized SwinYNet detector's 92.90%, while producing far fewer false positives o…
Visionaire MCP v0.7 introduces a verification layer that lets AI coding agents visually inspect live web pages, identify the exact CSS rule, file, and line causing rendering bugs, and verify fixes automatically. The tool…
French AI company Mistral released Robostral Navigate, a model that enables robots to follow plain-language instructions using a single RGB camera, achieving a 76.6% score on the R2R-CE benchmark. The model outperforms s…
French AI company Mistral has released Robostral Navigate, a new AI model that enables robots to navigate using only a single RGB camera and plain language instructions, achieving a 76.6% score on the R2R-CE benchmark. T…
A new study from Lancaster University finds that people are no better than chance at distinguishing AI-generated faces from real ones, and they actually perceive the fake faces as more trustworthy. This poses a serious r…
Elon Litman published a technical walkthrough of Joint Embedding Predictive Architectures (JEPA), Yann LeCun's self-supervised learning method that trains models to predict representations of masked image regions from vi…
Sony released the RX10 V bridge camera, reviving its all-in-one series with a 24-600mm ZEISS zoom lens and AI-powered autofocus for $2,299.99, shipping in August. The camera features a 20.1-megapixel sensor, 30 fps burst…
Roboflow integrated Workflows into Auto Label, enabling users to run custom multi-model pipelines serverlessly on unannotated images. The update allows combining models like Google Gemini, OpenAI GPT, and Anthropic Claud…
New research reveals that transformer-based vision-language models are vulnerable to adversarial attacks through their spectral structure, introducing the spectral-subspace-guided attack (SSGRA) that exploits these weakn…
Researchers have developed a reference-free evaluation framework for Vision-Language Models that converts flowchart images into code, using OCR and visual entailment to assess output quality without ground-truth data. Va…