A look into brain activity when we think, read, or listen.
― 8 min read
Computer Science - Computer Vision and Pattern Recognition
RSSA method to enhance low light images by reducing noise and improving contrast.
― 5 min read
OKO improves prediction accuracy and calibration by learning from sets of data points.
― 7 min read
New method improves identification of strange objects on the road.
― 7 min read
ETran efficiently ranks pre-trained models for object detection and image classification.
― 5 min read
A new framework combines neural networks with symbolic thinking for better understanding.
― 5 min read
RealCQA offers a fresh approach to tackling chart question answering challenges.
― 5 min read
Learn the process of retrieving data from various chart types.
― 6 min read
US2QNet improves ultrasound image quality assessment through an innovative unsupervised learning approach.
― 6 min read
Exploring how biases affect image models and strategies for mitigation.
― 5 min read
A new approach to accelerate semantic segmentation tasks in image processing.
― 6 min read
A new project enhances machine understanding of visual information through a large dataset.
― 5 min read
Improving object detection efficiency while maintaining simplicity.
― 5 min read
Improving base layer for better machine analysis and human viewing.
― 5 min read
A new method combines partial-label and zero-shot recognition for efficient image analysis.
― 6 min read
New methods improve object detection systems' resistance to adversarial changes.
― 5 min read
New methods create realistic human actions from text, improving gaming and film experiences.
― 8 min read
Discover the role of tensors in modern data analysis techniques.
― 7 min read
This method improves uncertainty estimates without needing model retraining.
― 6 min read
Combining LiDAR and camera data improves accuracy in 3D panoptic segmentation.
― 6 min read
A new AI method quickly generates creative book cover images.
― 5 min read
A novel method for tracking and segmentation using only appearance models.
― 7 min read
Our system improves AR by accurately estimating lighting and generating realistic shadows.
― 5 min read
This method enhances image classification through innovative class representation.
― 4 min read
Research connects specific neurons in models to interpreting images and generating text.
― 7 min read
MA-CLIP improves action recognition using few examples and better integration of video and text.
― 6 min read
Assessing how AI interprets images against human perception using a new dataset.
― 6 min read
A method for better image generation that safeguards privacy using public datasets.
― 6 min read
Learn how to keep search systems effective amid changing online content.
― 7 min read
Foundation models are enhancing efficiency in medical image segmentation.
― 7 min read
A new method uses GANs to create realistic radar data for safer automated driving.
― 4 min read
RadFM enhances radiology with multimodal data and advanced analysis capabilities.
― 5 min read
Introducing a model that predicts brain activity from images efficiently.
― 7 min read
Fourier-Net simplifies and speeds up medical image registration for better healthcare outcomes.
― 6 min read
MSECNet improves normal estimation accuracy and speed in 3D point clouds.
― 4 min read
A new method enhances 3D object detection for autonomous vehicles using BEV representation.
― 5 min read
New method uses video for sensor placement in tracking human activities.
― 4 min read
This method improves HDR images from single LDR photos using innovative technology.
― 5 min read
New methods improve image style transformation using Score-Decomposed Diffusion Models.
― 6 min read
SJTU-H3D provides essential tools for evaluating the quality of digital humans.
― 8 min read