Latest Articles for Computer Vision

New strategies improve image quality by effectively removing noise.

PartMix improves matching of visible and infrared images in person re-identification.

This research improves machine learning accuracy by selecting synthetic features effectively.

A new method improves 3D shape creation from 2D images using innovative techniques.

Examining how adversarial attacks impact segmentation models and exploring potential defenses.

A new approach to efficiently model and animate 3D surfaces using point clouds.

This project reveals how semantic segmentation improves accuracy in 3D modeling.

New method estimates 3D human poses using less labeled data.

A new method enhances image generation using fewer labeled images.

DPPD uses polygons for accurate object shape detection in real-time applications.

A new framework improves segmentation accuracy using limited labeled data.

A look at how learning invariance improves object recognition systems.

Introducing Receptive-Field Attention to enhance CNN performance in image tasks.

New methods aim to enhance how models link sounds to visuals.

A new approach to detect small objects efficiently on low-power devices.

This article explores new methods for enhancing medical image analysis using self-supervised learning.

A new method enhances the conversion of processed images to raw-like formats.

EGA-Depth improves depth estimation in autonomous vehicles using guided attention.

A new model simplifies image segmentation across various tasks.

Combining neural fields and meshes improves 3D scene creation for VR and AR.

Explore new methods in generative modeling for better data representation.

Researching how VQA models adapt to slight changes in questions and images.

GAPretrain improves accuracy in 3D object detection using camera and LiDAR data.

Combining 2D images and 3D data enhances segmentation accuracy in various applications.

This study investigates classifier performance with text-guided image alterations.

An analysis of how well visual-language models understand images and their meanings.

A new model improves camera depth and position estimation for various applications.

TeGRA improves camera tracking using event-based technology for enhanced accuracy.

SparseFormer mimics human focus to improve image and video recognition efficiency.

This approach improves image quality assessment by analyzing multiple images together.

Lift3D improves synthetic data generation for 3D vision tasks.

SQUAT improves scene graph generation by focusing on relevant object relationships.

A new model enhances infrared imaging for better 3D representation.

This article discusses enhancing model performance for less common categories in visual recognition.

A new model for identifying building shapes directly from images without extra processing.

This article details a novel approach to improve hazy image quality.

Examining Capsule Networks' strengths against traditional CNNs in robust image processing.

A new method enhances deep learning models against adversarial and natural changes.

New method enables robots to identify and segment objects autonomously.

This system improves mapping and tracking using dense neural point clouds.