Are Minimal Radial Distortion Solvers Really Necessary for Relative Pose Estimation? [0.03%]
最小径向畸变求解器真的有必要用于相对姿态估计吗？
Viktor Kocur,Charalambos Tzamos,Yaqing Ding et al.
Viktor Kocur et al.
Estimating the relative pose between two cameras is a fundamental step in many applications such as Structure-from-Motion. The common approach to relative pose estimation is to apply a minimal solver inside a RANSAC loop. Highly efficient s...
Structure-from-motion in micro-image domain for uncalibrated plenoptic 2.0 cameras [0.03%]
用于未校准光场2.0相机的微图像结构从运动法
Sarah Dury,Daniele Bonatto,Jaime Sancho et al.
Sarah Dury et al.
We introduce a structure-from-motion method specifically designed to process the raw micro-images captured by plenoptic 2.0 cameras. Unlike traditional monocular cameras, plenoptic cameras incorporate a micro-lens array between the sensor a...
FourierMIL: Fourier Filtering-based Multiple Instance Learning for Whole Slide Image Analysis [0.03%]
傅里叶滤波的基于整体滑动图像分析的多实例学习方法
Yi Zheng,Harsh Sharma,Margrit Betke et al.
Yi Zheng et al.
Recent advancements in computer vision, including convolutional neural networks, multilayer perceptrons, graph-based methods and transformer architectures, have significantly improved image classification. However, applying these techniques...
Nazir Nayal,Youssef Shoeb,Fatma Güney
Nazir Nayal
Addressing the Out-of-Distribution (OoD) segmentation task is a prerequisite for perception systems operating in an open-world environment. Large foundational models are frequently used in downstream tasks, however, their potential for OoD ...
From Forest to Zoo: Great Ape Behavior Recognition with ChimpBehave [0.03%]
从森林到动物园：使用ChimpBehave进行大猩猩行为识别
Michael Fuchs,Emilie Genty,Adrian Bangerter et al.
Michael Fuchs et al.
This paper addresses the significant challenge of recognizing behaviors in non-human primates, specifically focusing on chimpanzees. Automated behavior recognition is crucial for both conservation efforts and the advancement of behavioral r...
Jisui Huang,Ke Chen,Andreas Alpers et al.
Jisui Huang et al.
Existing level set models employ regularization based only on gradient information, 1D curvature or 2D curvature. For 3D image segmentation, however, an appropriate curvature-based regularization should involve a well-defined 3D curvature e...
A Physics-Informed Deep Learning Deformable Medical Image Registration Method Based on Neural ODEs [0.03%]
基于神经ODE的物理信息深度学习医学图像配准方法
Amirhossein Amiri-Hezaveh,Shelly Tan,Qing Deng et al.
Amirhossein Amiri-Hezaveh et al.
An unsupervised machine learning method is introduced to align medical images in the context of the large deformation elasticity coupled with growth and remodeling biophysics. The technique, which stems from the principle of minimum potenti...
Christen Millerdurai,Hiroyasu Akada,Jian Wang et al.
Christen Millerdurai et al.
Monocular egocentric 3D human motion capture remains a significant challenge, particularly under conditions of low lighting and fast movements, which are common in head-mounted device applications. Existing methods that rely on RGB cameras ...
A Closer Look at Benchmarking Self-supervised Pre-training with Image Classification [0.03%]
从图像分类的角度重新审视自监督预训练的基准测试工作
Markus Marks,Manuel Knott,Neehar Kondapaneni et al.
Markus Marks et al.
Self-supervised learning (SSL) is a machine learning approach where the data itself provides supervision, eliminating the need for external labels. The model is forced to learn about the data's inherent structure or context by solving a pre...
RNAS-CL: Robust Neural Architecture Search by Cross-Layer Knowledge Distillation [0.03%]
基于跨层知识蒸馏的鲁棒神经结构搜索(RNAS-CL)
Utkarsh Nath,Yancheng Wang,Pavan Turaga et al.
Utkarsh Nath et al.
Deep Neural Networks are often vulnerable to adversarial attacks. Neural Architecture Search (NAS), one of the tools for developing novel deep neural architectures, demonstrates superior performance in prediction accuracy in various machine...