Computer Vision news and analysis on Web Pulse: 4325 curated articles tracking the latest Computer Vision developments, tools, and research, updated continuously from vetted sources.
REFLEX, a new open-source tool, replays demonstrated GUI workflows with zero LLM calls per action, achieving 100 perfect runs out of 100 in a benchmark with randomized window positions and row orders, and a median replay…
Meta AI's DINOv2 (2023) scales the original DINO self-supervised vision model into a Vision Foundation Model that matches or beats text-guided models like CLIP on dense pixel-level tasks, introducing a patch-level object…
China has deployed an AI-driven fish facial recognition system to track migrating fish through a fishway on the Yarlung Tsangpo River in Tibet, replacing manual counts. The system, reported by Global Times, has helped ar…
NEC Corporation has developed a parking system that uses cameras and AI to detect when a driver exits their vehicle, starting the parking fee only at that moment. The technology, demonstrated at NEC's headquarters in Tok…
Two July 2026 papers demonstrate that AI models improve by compressing information rather than accumulating it. The Chinese Academy of Sciences' Institute of Automation released PhiZero, a world model that compresses vid…
A developer has created a real-time spine posture monitor using MediaPipe and Python that tracks body landmarks via webcam to detect slouching and send system notifications. The system calculates the angle between the ea…
A startup has developed a system that uses satellites and artificial intelligence to detect fires within minutes, enabling rapid response to wildfires. The technology leverages AI to analyze satellite imagery in near-rea…
A new conceptual architecture for a distributed drone-defense ecosystem combines multi-node sensor arrays, blockchain-governed drone coordination, and AI-driven multimodal decision-making. The system uses infrared signal…
NoRecognition, a project presented at Black Hat and DEF CON, is launching a limited drop of AI adversarial clothing designed to evade computer vision systems, with 50 pieces of each of four items (tee, hoodie, buff, and …
GaussianSplatting.jl 2.0, a Julia package for 3D Gaussian splatting, adds multi-GPU backend support (AMD, NVIDIA, MacBook), a multithreaded UI, MCMC densification, depth and geometry supervision, and a sky dome feature. …
Afrid Ali is developing EdgeAI Forge, an early-stage local-first architecture for using specialized AI agents in industrial vision, ROS 2, robotics, and edge-deployment workflows, and is seeking model recommendations and…
V2Fun, a browser-based AI platform developed by Zac Zuo, Tammy, and Yifu Wang, turns text, images, or video into rigged, textured 3D characters with automatic skeletons and AI motion capture, reaching number one Product …
Nissan Motor Co. has deployed AI-powered cameras at its Canton, Mississippi, assembly plant that monitor how workers bend, walk, and build cars, contributing to about $34 million in safety and health-related savings, acc…
Photosynthesis, a new iPhone camera app, captures simultaneous images from two lenses and fuses them into a single high-resolution photo using only real optical detail and no generative AI. The app, which supports 1x + z…
Intel's Core 5 320 'Wildcat Lake' processor, built on Intel 18A in Intel's own fabs, achieved 6.21 Gflops per watt in Jeff Geerling's HPL Linpack testing, outperforming Apple's A18 Pro in the $599 MacBook Neo at 5.38 Gfl…
Roboflow's Playground leaderboard ranks Qwen3.8-Max, Gemini 3.5 Flash, and other vision-language models for AI data labeling, where foundation models propose annotations that humans review to cut labeling time. The free …
Apple is expected to launch AirPods with cameras next month at its iPhone 18 Pro event, according to Mark Gurman's reporting, with the devices featuring visual intelligence that lets users ask Siri about their surroundin…
Roboflow published a tutorial for building an AI video car damage inspector that uses an RF-DETR model, ByteTrack tracking, and Gemini to generate digitally signed PDF reports. The system timestamps video fingerprints be…
Roboflow, a computer vision platform, has developed a deployment model that allows its cloud-native AI products to run on-premises within the Purdue Model's segmented industrial networks, supporting air-gapped, DMZ-broke…
Perforated AI released a perforated version of ResNet-18 on Hugging Face that matches ResNet-34's performance while maintaining ResNet-18's smaller size and compute efficiency, using dendritic optimization technology. Th…