Latest news, developments and reporting on Computer Vision.
World Labs released Atlas on Sept. 2, but Chinese startup Yingsu had already open-sourced a comparable world model in March and later ranked first in WorldArena 2.0. Yingsu also launched an open data initiative.
A Sun Yat-sen team's SpatialSV embeds explicit 3D geometry in MLLMs, improving spatial accuracy and halving manual annotation.
AI video chip SmarCo GC3 and 3D scene model WorldGen launch to ease content creation bottlenecks.
Dyson unveils $499 CameraJet, a smart toothbrush that uses a camera and on-device AI to detect gaps between teeth and automatically blast mouthwash into them.
FarseerTech, founded by ex-DJI engineers, is crowdfunding the RocXZoom, a robotic extreme zoom camera with 50x hybrid zoom and AI bird recognition.
Nvidia's DLSS 5 debuts with NBA 2K27 on Sep 3, RTX 50-series only; mixed early reviews cite realism and performance hit.
Dyson has launched the CameraJet, a camera-equipped electric toothbrush that uses AI-guided mouthwash jets to floss between teeth while brushing. The premium device, priced around $500 in the U.S., went on sale Tuesday.
Leiphone reports: TUM's Ziyue Li and a new Li Fei-Fei/Wu Jiajun paper both tackle AI's gap between benchmark success and real-world deployment.
A free, open-source tool called Paperwork helps Linux and Windows users tag, label, and search documents locally, according to ZDNet.
Amap, an Alibaba Group subsidiary, unveiled ABot-Recon, a streaming 3D reconstruction model that rebuilds 10,000-frame scenes in real time using just 12 local frames and no long-range memory.
OpenAI disclosed strategic missteps and a safety-focused reboot, Salesforce and Anthropic unveiled 'Claudeforce,' Alibaba launched Qoder, and Chinese firms pushed AI edge and industry applications.
Perceptron, founded by former Meta researchers, launched Isaac 0.5, an open-weight vision model to help robots perceive, reason and act in warehouses and factories.
OBSBOT, a Shenzhen AI camera maker, has become the global leader in high-end webcams with over 50% market share, according to co-founder Li Liang. The company sustained 50% annual growth and recently completed a Series D round.
Tsinghua and HKUST researchers propose LiveEdit, a streaming video editing framework achieving 12.66 FPS with 4-step denoising, enabling real-time editing for live scenarios. Accepted at ECCV 2026.
Underwater vehicles could navigate murky water using a sonar-guided technique from WHOI researchers, MIT Tech Review reports.
AI is not replacing radiologists, but it is changing their roles. Three-quarters of FDA-cleared AI medical devices target radiology, and AI tools are improving efficiency and accuracy in image interpretation.
Google Search has rolled out new features, including AI Mode, Lens, Circle to Search, Search Live, and price tracking, to help users with home decor projects as related searches surge.
Small e-ink displays and Apple's 1-megapixel AirPods camera show AI hardware is going low-res and ambient.
MarkTechPost details a deepDoctection 1.2.x workflow that combines layout detection, table recognition, OCR, reading-order reconstruction and JSONL export for RAG systems.
Ordnance Survey CEO Nick Bolton discusses transforming the 225-year-old institution into an AI powerhouse, emphasizing location data as a universal connector.
A TechRadar journalist tested the BodyPark Atom, an AI-powered home fitness coach, and was impressed by its movement mapping despite initial skepticism.
DeepSeek has introduced V4 Flash Vision Exp, a multimodal LLM that outperformed Anthropic's Opus 4.8 on two image benchmarks, with a paid platform launch and possible open-source release later.
SenseTime has open-sourced SenseNova U1.5 Lite, a lightweight multimodal model supporting 3-4k character instructions and native 4K output for stable visual creation.
DeepSeek unveils experimental vision model deepseek-v4-flash-vision-exp for agents, with Harness support and Files API.