Computer Vision news and analysis on Web Pulse: 3341 curated articles tracking the latest Computer Vision developments, tools, and research, updated continuously from vetted sources.
Kimi K3 now natively supports vision understanding, enabling coding agents to analyze UI screenshots and suggest changes. This shifts frontend workflows, requiring developers to handle image uploads for agents and ensure…
A developer released Enlarger, a local image upscaling tool designed to preserve detail rather than smoothing it, as an alternative to existing AI upscalers. The creator cited frustrations with heavy, complex, subscripti…
MIT is spending over $3 million on more than 500 AI surveillance cameras in academic buildings, residence halls, and outdoor areas along Memorial Drive, according to information obtained by The Tech. Installation began N…
Dubai International Airport has trialled a new AI-powered border control system that clears passengers in six to 14 seconds without requiring passports or boarding passes. The system, developed by Dubai Airports and the …
Dubai International Airport (DXB) has launched a biometric AI border control system that clears passengers in 3.4 seconds without requiring a passport or boarding pass, reducing terminal transit time by over 60%, accordi…
Google engineer Bhavuk Jain presented how modern AI systems become products, detailing two mobile features: AI wallpapers and Circle to Search. The process involves four architectural steps—post-training, fine-tuning, re…
A tutorial on building an edge AI safety helmet camera using YOLOv8 and the NE503 processor has been published, but the content is blocked by Cloudflare security. The site owner can be contacted for access.
Nvidia showed off more DLSS 5 footage during its SIGGRAPH 2026 keynote, detailing how the neural rendering technology preserves artistic intent through developer controls. Edward Liu, Nvidia's Director of Applied Deep Le…
A developer team tested the newly released open-source LTX Desktop app by Lightricks, which allows users to run the LTX-2.3 video model locally on their own computers with zero usage costs. The review compares it to Open…
A developer released an open-source tool that removes the Gemini sparkle watermark from AI-generated images by solving the alpha blend equation, recovering original pixels exactly without inpainting. The tool runs entire…
A user's field report of running the open-source tool claude-real-video on 2,181 videos revealed a critical design flaw: percentage-based frame deduplication was structurally blind to small subjects, causing it to delete…
Rhoda AI is betting the future of robotics starts with internet video, not lab data, using its Direct Video Action model that learns physics from hundreds of millions of clips. The company's headquarters visit revealed h…
NVIDIA released Cosmos 3 Edge, a 4-billion-parameter open world model designed to run on-device for robots and vision AI agents, enabling real-time reasoning and local action generation. The model is the smallest tier in…
NVIDIA showcased DLSS 5 neural rendering at SIGGRAPH 2026 in Los Angeles, revealing that game developers can choose from three models (Model A, Model B, Model C) with varying parameter counts for different levels of stru…
Activists in London altered a Meta billboard featuring Kylie Jenner to warn about potential secret surveillance from AI-powered smart glasses, sparking debate over privacy risks of wearable cameras and facial recognition…
Adobe is adding an experimental "AI Playground" feature to its Project Indigo iPhone camera app, powered by Google's Nano Banana model, that enables AI-powered photo editing, style transfers, and personalized photo guida…
Researchers introduced Retriever, a programming framework for closed-loop robot agents that supports modules running on independent clocks from milliseconds to seconds, and demonstrated Retriever-0, a bimanual manipulati…
Meta's smartglasses, which can record video covertly with only a small flashing light as a warning, pose a risk to children by enabling surreptitious filming that could feed AI-generated child abuse material, Guardian co…
China Telecom's Institute of Artificial Intelligence (TeleAI) and Harbin Institute of Technology have developed ShotPlan, a cinematic video generation method built on Wan2.1-T2V-14B that uses learnable planning tokens to…
Researchers introduced GenSyn10, a CIFAR-10-aligned synthetic image dataset of 60,000 images generated using three architecturally diverse models (FLUX.2-dev, HunyuanImage-3.0, and Qwen-Image-2512) to benchmark AI-genera…