Computer Vision

Shop Home Styles, a Fuel Ventures-backed platform, launched an AI-powered visual search tool that lets users upload photos to find matching furniture and home decor from over 100 brands. The tool analyzes image details l…

Unsupervised deep learning is transforming cardiac MRI reconstruction, with a new image-domain dual-branch INR framework (I-FP-INR) outperforming baseline methods. This approach eliminates the need for fully sampled refe…

TransitNet, a 1.5 MB deep-learning model, detects Earth-size planets with 95.2% accuracy in low-signal environments, outperforming traditional methods like TLS and BLS. It is 12-25 times faster than CPU-based TLS and has…

Researchers have developed a new sampling framework that optimizes consistency models to solve inverse imaging problems more efficiently, outperforming traditional diffusion models on metrics like FID and PSNR while requ…

A new AI framework called EGAgent uses entity scene graphs to achieve long-term video understanding, reaching a 57.5% performance rate on the EgoLifeQA dataset. The technology aims to enable wearable AI assistants to com…

Researchers at the University of Texas at Austin introduced VSeek, a framework that uses reinforcement learning to transform long-video question answering into an interactive retrieval process, improving Pass@1 scores by…

Researchers have developed a conditional Generative Adversarial Network (cGAN) framework that generates 3D porous media volumes with controlled porosity from 2D thin section images, eliminating the need for costly 3D tra…

Researchers have developed a new framework called Dual-Critic Guided Diffusion Alignment (DCDA) that enables autonomous vehicles to navigate adverse weather conditions without explicit weather modeling. By using a 4D rad…

SCALA, a new machine learning framework inspired by cognitive psychology, improves model accuracy and sample efficiency in data-scarce environments by guiding learning from broad categories to fine-grained recognition. T…

Researchers have developed End-Net, a convolutional neural network for multi-class MRI classification of neurological disorders, outperforming existing models in accuracy and generalization. The system addresses class im…

Researchers introduced Naïve PAINE, a method that improves text-to-image generation by pre-selecting initial noise based on user prompts to reduce randomness and enhance output quality. The approach outperforms existing …

Flow Matching models are outperforming diffusion-based methods in CT reconstruction, offering faster and more consistent results through deterministic ODEs. The FMCT and EFMCT frameworks reduce neural network evaluations…

A new Transformer-based model for sign language recognition achieves 88.4% accuracy on unseen classes from the LSA64 dataset using only eight reference examples per class, demonstrating strong one-shot learning capabilit…

Apple's EgoDex hand-manipulation dataset can now be queried using natural language and geometric features via Daft, an open-source data framework. The approach combines SigLIP semantic embeddings with hand-pose geometry …

In a head-to-head comparison of AI video generation models, Gemini Omni Flash decisively defeated AnimateDiff with a score of 34.8 to 14.6, winning all four tasks with 97% confidence. Gemini Omni Flash outperformed on pr…

Path Robotics CEO Andy Lonsberry discussed on The Robot Report Podcast how the company uses AI and real-time vision guidance to optimize robotic welding, including deploying Boston Dynamics' Spot robots for shipbuilding.…

Researchers introduced SFT+RL, a two-stage framework combining supervised fine-tuning and reinforcement learning on CLIP's encoder, achieving a 10.2% improvement in clean accuracy and 15.8% in adversarial robustness on U…

Researchers developed an AI framework using semantic segmentation on H&E-stained images to diagnose liver cancer types, achieving high accuracy rates of up to 1.000 for colorectal metastatic adenocarcinoma. The method co…

Vision-language models exhibit a source-credibility bias, prioritizing logos and names over content, according to a study by the CueTrust initiative. The bias, measured via a Source-Override Index, was found across seven…

PixCon, a new framework for semi-supervised semantic segmentation, guarantees contamination-free pseudo-labeling by using a per-class memory bank that only admits correctly classified labeled pixels. It outperforms or ma…