Tag: #vision-ai14 results
NVIDIA DeepStream 9 uses coding agents like Claude Code or Cursor to simplify building real-time vision AI applications. It eliminates complex data pipelines and lengthy code development. Developers can now create deployable, optimized vision AI code easily.
Tesla's regulated FSD version received approval from Dutch regulator RDW for supervised use, marking its European debut. The system manages steering, acceleration, and braking on highways and city roads after 18 months of testing. Deployment starts in Netherlands, with EU-wide rollout pending vote.
NVIDIA introduces Batch Mode VC-6 to accelerate vision AI pipelines by matching decode, preprocessing, and GPU scheduling to improving model throughput. It addresses the data-to-tensor gap from prior CUDA-accelerated VC-6 posts. NVIDIA Nsight aids in optimizing these pipeline stages.
Google updated Circle to Search with Gemini 3 to detect and search multiple objects in images simultaneously. It enables shopping matches for outfits including accessories with virtual try-on, and reasons object relationships like fish coexistence in reefs. First rollout to Galaxy S26 and Pixel 10 devices.
World Labs, startup by AI pioneer Fei-Fei Li, secured $1 billion in funding. The capital will advance a novel approach to AI development.
QbitAI tests Xunfei Zhiwen Vision Agent, claiming it generates perfect PPTs without rework. The tool uses AI vision for seamless presentation creation. Ideal for eliminating manual edits in slide decks.
Elon Musk conceded that Tesla's HW3 hardware lacks the capability for unsupervised Full Self-Driving due to insufficient memory bandwidth, delaying rollout by years. This impacts 4 million HW3 owners, offering limited compensation like HW4 upgrades or discounted new cars. Tesla's pure-vision approach faces criticism amid Waymo's sensor-fusion success.
JD's JoyInside platform powers innovative AI hardware like animating children's drawings, magnifier-based object recognition, and soul-infused desk lamps. It positions as super infrastructure for the AI hardware era, bringing sci-fi concepts to reality.
University of Washington team unveiled VueBuds, TWS earbuds with embedded micro cameras and VLM for real-time voice descriptions of the wearer's surroundings. It identifies objects, translates text, and enables voice interaction. Described as a mobile, voice-only reverse image search prototype.
Shenzhen-based LDRobot refiled for Hong Kong IPO after CSRC approval. It reported $52.9 million revenue in H1 2025 while narrowing losses. The firm claims to be the world's largest visual perception-focused robotics company.