Latest Articles for Computer Vision

i-SRN improves pose estimation for robots using implicit representations and neural rendering.

Exploring methods to improve image coding for advanced AI applications.

A technique to identify unreliability in human body mesh reconstruction.

VoxDet improves object recognition by using 3D models to tackle complex scenes.

GRAtt enhances tracking efficiency in challenging video segmentation tasks.

Exploring how neural networks recognize symmetries in data through equivariance.

New methods in knowledge distillation enhance model training efficiency.

Analyzing limitations and corrections in influence functions for better model performance.

This study enhances a classic method for detecting lines in document images.

PlaNeRF enhances 3D modeling from 2D images, improving geometry and image quality.

A new method improves data sampling using normalizing flows and Langevin dynamics.

New methods improve machine learning models' ability to handle unseen data.

Research integrates biological principles into CNNs for better image analysis.

Introducing a modular method for zero-shot visual question answering.

A new method aims to enhance object localization accuracy in video analysis.

This method improves how computers connect images with captions.

A new method improves image quality using limited high-resolution data.

A new dataset enhances scene graph parsing for better image and text connections.

GMSF offers a fresh approach to estimating motion in 3D space.

New methods improve 3D reconstruction of reflective surfaces using neural rendering techniques.

This article investigates the necessity of the query component in transformer models.

A new model enhances data generation from multiple input types.

T2FNorm improves neural networks' ability to detect unfamiliar data.

Learn about YOLO for real-time object detection.

Learn to create a system that identifies vehicle wheels in varying conditions.

Researchers use images to teach robots how to interact with the world.

New framework improves accuracy of 3D object localization using a single camera.

This study examines how deep learning models interpret logic in diagrams using visual illusions.

New dataset enhances image-question capability in Hausa language processing.

Caterpillar is a novel MLP architecture for capturing local image details.

A new approach for running X3D model on FPGAs for efficient video analysis.

A framework to enhance generative models using pre-trained diffusion models.

A new approach integrates kernel methods with deep learning for better performance.

A new method enhances vision-language models through real-time feedback for better performance.

LayoutMask enhances text and layout interaction for better document comprehension.

A new approach to improve scene graph generation for better visual understanding.

PaLI-X combines vision and language skills, excelling in diverse tasks.

This study assesses different techniques for detecting 3D shapes under rotation.

New neural networks learn transformations directly from data, improving efficiency and understanding of symmetries.

SlimFit reduces memory use for transformer models during fine-tuning.