Sunny Optical is extending its Optics Plus AI strategy into humanoid robots and smart glasses, showing vision modules, 3D sensing and lightweight AR optics at CIOE. It says robot vision and smart glasses need clearer, lighter optical systems, while its phone-era manufacturing base supports system-level delivery.
TechRadar reports that Linux developer Justin Schroeder demonstrated an AI coding agent using an older Intel MacBook, webcam and mirror to visually inspect AMD Radeon driver changes in Omarchy Linux.
iOS 27 updates the iPhone Camera app with a new Siri mode for AI-compatible models, customizable controls and exclusive iPhone 18 Pro tools, according to 9to5Mac.
Former TikTok employees Melody Chu and Jing Liu have launched Superpose, an iOS camera app that uses AI to suggest poses. It has been downloaded more than 22,000 times since July.
Buildots Ltd. has raised $130 million, bringing total funding to $297 million, to expand its computer-vision construction platform across North America and Europe and extend it through more stages of construction.
Local governments across the U.S. are canceling Flock license plate camera contracts amid privacy concerns, even as President Trump backs the technology, a new poll shows public opposition, and police misuse cases mount.
Companies · Products & Applications · Large Language Models · Robotics · Computer Vision
DeepCybo released PhysBrain 1.5, an open physical AI model scoring 72.5 on 28 benchmarks, while ECCV 2026 presentations from TUM and NUS Show Lab and CosmosMind's MetaRSI-v1 advanced spatial intelligence, unified multimodal models and recursive self-improvement.
NVIDIA's DLSS 5, announced in March and leaked in August, is drawing criticism from gamers and developers who say its AI rendering changes game art direction. NVIDIA says developers retain full control.
9to5Mac outlines how older iPhones can create background blur using Portrait mode, Macro mode, close focusing and telephoto lenses, even as the iPhone 18 Pro adds a wider f/1.48 variable aperture.
Stanford's Jiajun Wu told ECCV 2026 that vision, hearing and touch are projections of one set of physical properties onto different channels, and that fusing them under scarce data calls for physical principles rather than bigger architectures. He described distilling stiffness from video diffusion models and turning a single image into an interactive 3D scene.
Companies · Large Language Models · Computer Vision
Amap has fully AI-enabled its 2026 Street Scan rankings and introduced ABot-Earth 0.7, a 3D-native city world model that generates realistic 3D scenes from satellite images or text. It is extending spatial intelligence from navigation into robotics and physical-world infrastructure.
Amazon is adding Prime Video shopping features that let viewers find products tied to what they are watching through X-Ray, Amazon Lens, and an expanded Shop the Show.
Feyn released MultiMatte, an open-source image background removal model that can be prompted with words to keep a named object and remove everything else. It is built on Meta's SAM 3 and uses alpha mattes.
ECCV 2026 opened in Malmö with record submissions, a surge of world-model workshops, and Chinese teams advancing 3D Gaussian Splatting and embodied AI.
At ECCV 2026 in Malmö, Tec-Do Technology led the MARS2 Workshop on multimodal reasoning and slow thinking, the only Agentic Commerce workshop initiated by a Chinese tech company. Its challenge drew 64 teams and more than 1,060 submissions.
Large Language Models · Robotics · Computer Vision
On September 10, 2026, QbitAI reported that Ant Lingbo released a 1.3B real-time world model, Amap launched a 3D-native city world model covering 196 countries, and researchers introduced HSImul3R to turn human videos into robot skills.
OpenAI released ChatGPT Images 2.5 on Sept. 9, adding precise editing, Sketch references and new API models. Weekly image generations reportedly exceed 3 billion.