Computer Vision news and analysis on Web Pulse: 1951 curated articles tracking the latest Computer Vision developments, tools, and research, updated continuously from vetted sources.
Researchers at arXiv have identified that Batch Normalization causes gradient skew in dynamic sparse training (DST) methods, leading to slower convergence compared to dense neural network training. The team proposed Spar…
Researchers have developed AGFlow, a spatiotemporal flow-matching model that fuses asynchronous Sentinel-1 radar and Sentinel-2 optical satellite data to remove clouds and reconstruct missing time-series frames. The mode…
Researchers have developed STROP, a discrete visual tokenizer that learns to generate variable-length structural scene representations, determining the optimal sequence length for each image in a single forward pass. The…
Researchers have introduced Tensor Memory, a lightweight module that gives Transformer models a fixed-size recurrent 3D memory tensor to handle long video sequences. The module writes tokens into a voxel grid using a dif…
A study evaluating FP4 quantization-aware training (QAT) for real-time anomaly segmentation in brain tumor detection found that architecture choice, not QAT recipe, most determines model quality. The Swin Transformer dem…
A new study from arXiv reveals that multimodal large language models (LLMs) frequently produce hallucinated outputs in agricultural imaging tasks, generating biologically inconsistent or agronomically implausible results…
Researchers have developed ForestHG-Trace, a framework enabling traceable long-horizon ecological reasoning over large-scale forest scenes by representing multimodal NEON forest data as ecological hypergraphs. The system…
A preliminary clinical validation of the Melanoscope AI mobile dermoscopy clinical decision support system, conducted across four "Melanoma Day" screening sessions in Orel, Russia from June 2025 to April 2026, found no f…
Researchers have developed representation-conditioned diffusion models that generate synthetic training images by conditioning on learned representations from DINOv2, DINOv3, and CLIP, achieving a 10.76 percentage point …
The 10th Affective & Behavior Analysis in-the-Wild (ABAW) Workshop and Competition, held at CVPR 2026, introduced new challenges for multimodal human-centered AI, including emotional mimicry intensity estimation, ambival…
Researchers at the University of Cambridge have developed the first large-scale Ukrainian handwritten text generation system, creating a dataset of 126,177 images from 308 writers and retraining a diffusion-based model t…
A design enthusiast highlighted the creative use of visual metaphors, such as a butterfly combined with a jumper, and called on AI researchers to develop evaluations for symbolic thinking in the visual domain. The post c…
Google has introduced a storyboard-driven video creation workflow using Gemini Omni, allowing users to control camera angles, terrain, and character swaps by feeding the model image sequences and timestamp prompts. The m…
Apple is presenting new research at the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026, taking place in person at the Colorado Convention Center in Denver from June 3 to June 7. The company is…
Google Flow, Google's AI filmmaking platform built on Veo 3 and Gemini, launched at Google I/O 2025, offering structured control over scenes, camera angles, and character consistency for video generation and editing. The…
Google unveiled Gemini Omni at Google I/O 2026, a multimodal AI model designed for video editing, compositing, and remixing rather than just generating clips from scratch. The model powers Google Flow's advanced editing …
Uvilox AI launched a real-time sign language interpretation platform that translates American and British Sign Language into text and voice with 97.4% accuracy and sub-80 millisecond latency. The system integrates direct…
Artificial intelligence and telemedicine are transforming ophthalmology patient journeys in the United Kingdom and Germany, with AI medical devices increasingly integrated into routine ophthalmic care. The shift is affec…
A user is seeking a generative AI model capable of first designing consistent characters—such as an ogre, mage, or dragonborn—based on an RPG campaign, and then using those designs to generate comic panels from story not…
Elodin has released an open-source AI racing simulation harness designed to match the constraints and message format of the upcoming AI Grand Prix virtual qualifier. The tool runs against real Betaflight firmware and gen…