Single-image method can accurately evaluate fibrous material porosity

Researchers from Skoltech Engineering Center's Hierarchically Structured Materials Laboratory have developed a new method to ...
eLife

Human EEG and artificial neural networks reveal disentangled representations and processing timelines of object real-world size and depth in natural images

Neural and computational evidence reveals that real-world size is a temporally late, semantically grounded, and hierarchically stable dimension of object representation in both human brains and ...
IEEE

Real-Time Object Detection: A Comparative Analysis of YOLO, SSD, and EfficientDet Algorithms

Abstract: With recent technological developments, object detection has become an important component of many computer vision applications, such as autonomous driving, image recognition, unmanned ...
IEEE

Boltzmann Attention Sampling for Image Analysis with Small Objects

Abstract: Detecting and segmenting small objects, such as lung nodules and tumor lesions, remains a critical challenge in image analysis. These objects often occupy less than 0.1% of an image, making ...