Results 61 to 70 of about 26,718 (222)

PatchZero: Defending against Adversarial Patch Attacks by Detecting and Zeroing the Patch

2023 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2023
Accepted to WACV ...
Ke Xu, Yao Xiao, Zhaoheng Zheng, Kaijie Cai, Ram Nevatia   +4 more
openaire   +2 more sources

Cell Segmentation Beyond 2D—A Review of the State‐of‐the‐Art

Advanced Intelligent Discovery, EarlyView.
Cell segmentation underpins many biological image analysis tasks, yet most deep learning methods remain limited to 2D despite the inherently 3D nature of cellular processes. This review surveys segmentation approaches beyond 2D, comparing 2.5D and fully 3D methods, analyzing 31 models and 32 volumetric datasets, and introducing a unified reference ...
Fabian Schmeisser, Haseeb Muhammad, Junaid Younas, Gillian Lovell, Bianca Migliori, Andreas Dengel, Sheraz Ahmed   +6 more
wiley   +1 more source

Improving Unsupervised Defect Segmentation by Applying Structural Similarity to Autoencoders

, 2019
Convolutional autoencoders have emerged as popular methods for unsupervised defect segmentation on image data. Most commonly, this task is performed by thresholding a pixel-wise reconstruction error based on an $\ell^p$ distance. This procedure, however,
Bergmann, Paul, Fauser, Michael, Löwe, Sindy, Sattlegger, David, Steger, Carsten   +4 more
core   +1 more source

Perceptual-Sensitive GAN for Generating Adversarial Patches

Proceedings of the AAAI Conference on Artificial Intelligence, 2019
Deep neural networks (DNNs) are vulnerable to adversarial examples where inputs with imperceptible perturbations mislead DNNs to incorrect results. Recently, adversarial patch, with noise confined to a small and localized patch, emerged for its easy accessibility in real-world.
Aishan Liu, Xianglong Liu 0001, Jiaxin Fan, Yuqing Ma, Anlan Zhang, Huiyuan Xie, Dacheng Tao   +6 more
openaire   +2 more sources

Uncertainty‐Guided Selective Adaptation Enables Cross‐Platform Predictive Fluorescence Microscopy

Advanced Intelligent Discovery, EarlyView.
Deep learning models often fail when transferred to new microscopes. A novel framework overcomes this by selectively adapting the early layers governing low‐level image statistics, while freezing deep layers that encode morphology. This uncertainty‐guided approach enables robust, label‐free virtual staining across diverse systems, democratizing ...
Kai‐Wen K. Yang, Andrew Bai, Alexandra Bermudez, Yunqi Hong, Zoe Latham, Iris Sloan, Michael Liu, Vishrut Goyal, Cho‐Jui Hsieh, Neil Y. C. Lin   +9 more
wiley   +1 more source

Deep Generative Adversarial Compression Artifact Removal

, 2017
Compression artifacts arise in images whenever a lossy compression algorithm is applied. These artifacts eliminate details present in the original image, or add noise and small structures; because of these effects they make images less pleasant for the ...
Bertini, Marco, Del Bimbo, Alberto, Galteri, Leonardo, Seidenari, Lorenzo   +3 more
core   +1 more source

BMPCQA: Bioinspired Metaverse Point Cloud Quality Assessment Based on Large Multimodal Models

Advanced Intelligent Systems, EarlyView.
This study presents a bioinspired metaverse point cloud quality assessment metric, which simulates the human visual evaluation process to perform the point cloud quality assessment task. It first extracts rendering projection video features, normal image features, and point cloud patch features, which are then fed into a large multimodal model to ...
Huiyu Duan, Kang Fu, Sijing Wu, Yunhao Li, Zicheng Zhang, Qiang Hu, Xiongkuo Min, Guangtao Zhai   +7 more
wiley   +1 more source

Adversarial Patch Attacks on Monocular Depth Estimation Networks

IEEE Access, 2020
Thanks to the excellent learning capability of deep convolutional neural networks (CNN), monocular depth estimation using CNNs has achieved great success in recent years.
Koichiro Yamanaka, Ryutaroh Matsumoto, Keita Takahashi, Toshiaki Fujii   +3 more
doaj   +1 more source

Calibration‐Free Electromyography Motor Intent Decoding Using Large‐Scale Supervised Pretraining

Advanced Intelligent Systems, EarlyView.
Calibration‐free electromyography motor intent decoding is enabled through large‐scale supervised pretraining across heterogeneous datasets. A Spatially Aware Feature‐learning Transformer processes variable channel counts and electrode geometries, allowing transfer across users and recording setups. On a held‐out benchmark, fine‐tuned cross‐user models
Alexander E. Olsson, Nebojša Maleševic, Anders Björkman, Christian Antfolk   +3 more
wiley   +1 more source

IPG: Incremental Patch Generation for Generalized Adversarial Patch Training

CoRR
The advent of adversarial patches poses a significant challenge to the robustness of AI models, particularly in the domain of computer vision tasks such as object detection. In contradistinction to traditional adversarial examples, these patches target specific regions of an image, resulting in the malfunction of AI models.
Wonho Lee, Hyunsik Na, Jisu Lee, Daeseon Choi   +3 more
openaire   +2 more sources