Exploration

Researchers at the intersection of computer vision and medical imaging have cracked a persistent problem in CT scan analysis: distinguishing not just whether a pathology exists, but how severe it is. Current vision-language models trained on CT scans

A team of researchers has developed an AI system capable of diagnosing congenital uterine anomalies from 3D ultrasound scans with accuracy that rivals experienced sonographers. The system, named CUA-Net, uses a modified 3D ResNet-18 architecture—an a

Thermal drones are quietly rewriting the rules for locating buried ordnance, and a new study from Tennessee offers the clearest picture yet of how this technology could reshape mine action. Researchers Chad Melton and Annabelle Kelton spent four fiel

Vision Transformers have quietly rewritten the rules of computer vision, turning raw pixels into structured embeddings with a fluency that rivals human intuition. Yet their greatest strength—their ability to generalize across semantic classes—has als

A new framework from a team of researchers at Tsinghua University and the University of Sydney solves a long-standing problem in machine vision: when a generative AI model produces an image that looks correct to a human but cannot be retrieved by the

A team of European researchers has developed a machine learning method to detect and map Christmas tree plantations using high-resolution aerial imagery, a breakthrough that could reshape how governments and industries monitor agricultural land use—e

Researchers have taken a significant step toward solving a persistent problem in satellite imagery analysis: adapting advanced vision-language models like CLIP to the unique challenges of remote sensing without drowning in the need for labeled data.