Computer Vision news and analysis on Web Pulse: 4315 curated articles tracking the latest Computer Vision developments, tools, and research, updated continuously from vetted sources.
James Hurley gave a talk on real-time video upscaling using convolutional neural networks and WebGPU. The presentation focused on performance considerations for running CNN-based upscaling in the browser.
Apple's visionOS 27 and iOS 27 bring the first object-tracking update since visionOS 2, adding high frame rate tracking, an extended training mode in Create ML, a metric pose, faster-trained .referenceobject files, and o…
The Department of Homeland Security has declined to classify its AI-driven immigration enforcement tools as high-impact, exempting them from the disclosure, impact-assessment, and appeal safeguards required under the Tru…
Eight supply chain leaders told Business Insider that AI's biggest practical gains are in data intake and gate automation, with Arrive Logistics chief strategy officer J-Ann Toles saying AI converts customer emails and b…
Researchers submitted MintAct, a family of vision-language models trained at 2B, 4B, and 8B scales that unifies UI grounding, multi-step navigation across mobile, desktop, and web, and visual tool use, to arXiv on 18 Sep…
Akhouri Systems, in partnership with the HYNAWEB ecosystem, has launched APIC (Advanced Image Processing Center), a browser-based image editing suite that runs entirely on-device using WebAssembly and on-device AI so tha…
A research paper titled "Refinement Is Inherently Editable: Training-Free Prompt-to-Prompt Image Editing with Generative Refinement Network" proposes a training-free approach to text-guided image editing that avoids the …
Legend.org, a New York-based AI company that purchased the nonprofit Mastery Transcript Consortium in July, is developing AI tools to give teachers and students immediate feedback on papers, exams, recorded presentations…
Gemini 3.8 Flash identified spider species correctly on 997 of 2,000 photos, a 49.85% accuracy rate that was the highest among nine AI models tested in a benchmark built from research-grade iNaturalist observations cover…
A developer benchmarked whether large language models can judge UI style-rule compliance from screenshots by comparing GPT and Gemini verdicts against DOM-derived gold labels across 60 contrastive rule pairs. GPT scored …
A community member is seeking guidance on dataset sources and training architecture for a project that empties indoor spaces and removes specific objects from images using provided object masks. The poster asks whether p…
A developer has introduced YOalphabet and YOconlang, a spatial 4-bit bytecode script and constructed language that maps visual glyph geometry directly to binary hardware registers, eliminating the need for vector embeddi…
Alibaba's Qwen team released Qwen-Image-2.1, an open-source image model that unifies text-to-image generation and image editing in a single workflow, with a 7 billion parameter visual generation component. Qwen-Image-2.1…
Pinterest reported an 18 percent lift in visits after deploying an AI system that generates five to seven likely searches per image and links them into collection pages, according to a February 2026 paper by Pinterest re…
Researchers introduced curvature-adaptive tubular correction (CAT), a training-free plugin that regulates gradient-guided diffusion sampling by decomposing the guidance gradient into normal and tangent components and cha…
A three-stage parameter-efficient framework adapting DINOv3 Vision Transformer models to synthetic aperture sonar (SAS) automatic target recognition found that Low-Rank Adaptation (LoRA) alone raised area under the preci…
A multimodal reasoning framework using Task-Aware Group Relative Policy Optimization (GRPO) achieved an Overall Score of 83.24 on the official PCBA Standard-to-Real Grand Challenge leaderboard, according to arXiv paper 2…
Researchers introduced MegaEvent, an event-based visual place recognition system that achieves an average Recall@1 of 82% across three existing event-based localization datasets, leading the next best event-based method …
A new arXiv paper (2609.21849v1) presents three contributions for running diffusion pipelines on consumer GPUs: an embedding translator that maps a small text encoder into a large encoder space to cut weight and latency,…
A new arXiv paper (2609.20839v1) proposes progressive error curriculum training (PECT), a curriculum learning framework that adapts a No Language Left Behind (NLLB)-based phoneme-to-text reconstruction model using synthe…