arXiv cs.CV
AI news source. Articles are auto-selected and adapted by Hamidun News editors.
Latest publications
Flow matching: velocity scaling cuts FID from 28.0 to 12.2 on ImageNet-256
An arXiv paper attributes the velocity-scaling effect in flow matching to a population time lag, not to underestimated velocity. FID on ImageNet-256 fell from 28.0 to 12.2 at 25 steps.
VICO Trains VLMs on Rewritable Images: Up to +5.0% on New Tasks
The VICO framework jointly trains a VLM actor and an image editor: task difficulty adapts to the model, and no annotation is needed. Out-of-domain gains reach +5.0%.
DOGS Beats 9 Baseline Methods at Logo Generation Across Two Renderers
DOGS drops prompt rewriting: a GFlowNet samples a logo design graph and penalizes similarity to existing marks. The authors compared it with nine baseline methods.
MESSENGER cuts 3D motion forecasting error by 71.6% on nuScenes
Researchers propose MESSENGER: scene flow estimation with a memory buffer and autoregression. On nuScenes, long-term EPE3D error fell 71.6%; on Argoverse 2, 67.7%.
LinSlot Teaches a Neural Network to Break a Scene Into Objects and Their Attributes Without Labels
The new LinSlot architecture finds objects and their attributes in images without labels, building on the linear representation hypothesis. The authors claim DCI gains on several datasets.
OcuFlow Removes Eyeglass Glare in One Step and Wins 67.32% of Votes in a Blind Test
The OcuFlow authors built OcuBench, a benchmark of 11 470 images, and trained a model that removes eyeglass glare in a single step. In a blind survey it beat its competitors by a factor of 6.2.
Researchers release VR180 dataset shot on Blackmagic cameras: 1211 stereo samples
First VR180 dataset captured directly on Blackmagic URSA Cine Immersive: 636 videos, 575 photos, factory calibration, and a finding that fisheye needs 38% more bitrate.
SLVR Adds 14.2 Points on BLINK Relation to Qwen2.5-VL-7B Without Text CoT
Researchers propose SLVR: latent reasoning with typed stages. On Qwen2.5-VL-7B, the method added +9.4 on MMVP and +14.2 on BLINK Relation without generating reasoning chains.