Article
Robotics
Issa Mouawad, Nikolas Brasch, Fabian Manhardt, Federico Tombari, Francesca Odone
Summary: Monocular 3D object detection is cost-effective and widely available, but annotation complexity limits dataset size. This research proposes a self-supervised method that uses temporal consistency of object poses as a supervision signal to improve detection performance by refining pose predictions and generating high-quality pseudo labels.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Chemistry, Analytical
Shishir Muralidhara, Khurram Azeem Hashmi, Alain Pagani, Marcus Liwicki, Didier Stricker, Muhammad Zeshan Afzal
Summary: This paper proposes an attention-heavy framework for video object detection, which aggregates disentangled features extracted from individual frames to tackle challenges introduced by video data. The proposed approach demonstrates significant performance improvement on the ImageNet VID dataset.
Article
Biology
Stijn Adriaan Nuiten, Andres Canales-Johnson, Lola Beerendonk, Nutsa Nanuashvili, Johannes Jacobus Fahrenfort, Tristan Bekinschtein, Simon van Gaal
Summary: Humans can detect conflict automatically even in the absence of conflict awareness, but neural signatures of conflict are only present when at least one feature of a potentially conflicting stimulus is attended, indicating an attentional bottleneck at the level of objects for cognitive control operations involved in conflict detection.
Article
Multidisciplinary Sciences
Omisa Jinsi, Margaret M. Henderson, Michael J. Tarr
Summary: Poor visual acuity in newborns may have functional advantages, allowing for faster and more accurate learning of basic-level visual object categories, which can also transfer to subordinate-level category learning.
Article
Computer Science, Artificial Intelligence
Jing Liu, Jiaxiang Wang, Zhiwei Fan, Min Yuan, Weikang Wang, Jiexiao Yu
Summary: This paper proposes a novel group attention retention network (GARNet) for co-salient object detection. By designing attention modules and retention modules, this method can better construct collaborative relationships across images and retain key features. Experiments show that GARNet outperforms previous methods on four datasets.
MACHINE VISION AND APPLICATIONS
(2023)
Article
Nanoscience & Nanotechnology
Zhenhan Zhang, Shuiyuan Wang, Chunsen Liu, Runzhang Xie, Weida Hu, Peng Zhou
Summary: This study introduces a hardware device inspired by the retina, with all-in-one perception, memory, and computing capabilities for the detection and recognition of moving objects. It achieves fast and efficient detection and recognition, outperforming previous results.
NATURE NANOTECHNOLOGY
(2022)
Article
Computer Science, Information Systems
Sadia Afroze, Md. Rajib Hossain, Mohammed Moshiul Hoque
Summary: This paper proposes a vision-based framework using head poses to detect the focus of attention in humans. Experimental results show that the proposed model outperforms other deep learning models and achieves high accuracy in multi-object scenarios.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2022)
Article
Robotics
Peixuan Li, Huaici Zhao
Summary: In this work, the novel one-stage and keypoint-based framework KM3D-Net is proposed for monocular 3D object detection using only RGB images, achieving superior efficiency and accuracy on the popular KITTI dataset. Additionally, this is the first successful application of semi-supervised learning in monocular 3D object detection, surpassing most previous fully supervised methods with minimal labeled data.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Chemistry, Analytical
Tianming Xie, Zhonghao Zhang, Jing Tian, Lihong Ma
Summary: In this paper, a novel target-aware token design for transformer-based object detection is proposed. The proposed method introduces a target-aware sampling module and a target-aware key-value matrix to tackle the target attribute diffusion challenge. Experimental results show that the proposed Focal DETR achieves superior performance over existing models on the COCO object-detection benchmark dataset.
Article
Computer Science, Hardware & Architecture
Naiyuan Chen, Yan Li, Zhuomin Yang, Zhensong Lu, Sai Wang, Junang Wang
Summary: This paper proposes a lightweight object detection network in UAV vision (LODNU) based on YOLOv4, which can meet the application requirements of resource-constrained devices while ensuring the detection accuracy. LODNU reduces the model parameters by using depth-wise separable convolution to reconstruct the backbone network and improves the extraction ability of key object features by embedding improved coordinate attention. The experimental results show that LODNU achieves performance near to YOLOv4 with significantly fewer parameters and floating-point operations.
JOURNAL OF SUPERCOMPUTING
(2023)
Review
Psychology, Mathematical
Patrick Cavanagh, Gideon P. Caplovitz, Taissa K. Lytchenko, Marvin R. Maechler, Peter U. Tse, David L. Sheinberg
Summary: The allocation of attention to objects and its mechanisms are explored in this review. Findings suggest that preattentive targets of object-based attention can be fully developed object representations. Control of object-based attention is believed to come from ventral visual areas that specialize in object analysis. The relationship between different modes of attention and object-based attention is also discussed.
PSYCHONOMIC BULLETIN & REVIEW
(2023)
Article
Robotics
Lin Zhao, Meiling Wang, Yufeng Yue
Summary: Camera-LiDAR fusion is a promising option for 3D vehicle detection in autonomous driving scenarios. Traditional camera-LiDAR based methods may perform poorly when lacking semantic segmentation labels. To address this issue, we propose a novel semantic augmentation method that improves the performance of multimodal 3D vehicle detection and achieves significant improvements in challenging detection scenarios.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Robotics
Andras Palffy, Ewoud Pool, Srimannarayana Baratam, Julian F. P. Kooij, Dariu M. Gavrila
Summary: In this study, a state-of-the-art object detector (PointPillars) is applied to 3+1D radar data. The benefits of additional elevation information, together with Doppler, radar cross section, and temporal accumulation, are explored in multi-class road user detection. Results show that object detection performance on 64-layer LiDAR data outperforms that on 3+1D radar data, but the gap can be reduced by adding elevation information and integrating successive radar scans. The VoD dataset, a valuable experimental dataset, is provided for scientific benchmarking.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Robotics
Azarakhsh Keipour, Guilherme A. S. Pereira, Sebastian Scherer
Summary: The proposed algorithm is suitable for real-world robotics applications, lightweight, and capable of handling lighting variations and partial view for elliptic pattern detection. Performance is demonstrated in real-world robotics task, outperforming other methods.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Environmental Sciences
Di Tian, Yi Han, Yongtao Liu, Jiabo Li, Ping Zhang, Ming Liu
Summary: This paper introduces a hybrid cross-feature interaction (HCFI) attention module for object detection in intelligent mobile scenes, which demonstrates superior performance through experiments.
Review
Neurosciences
Jiye G. Kim, Emma Gregory, Barbara Landau, Michael McCloskey, Nicholas B. Turk-Browne, Sabine Kastner
PROGRESS IN NEUROBIOLOGY
(2020)
Review
Neurosciences
Sabine Kastner, Ian C. Fiebelkorn, Manoj K. Eradath
CURRENT OPINION IN NEUROBIOLOGY
(2020)
Article
Psychology, Biological
Timo Stein, Marius Peelen
Summary: Stein and Peelen demonstrate that conscious and unconscious contributions to detection effects in face perception and attention tasks can be distinguished using discrimination performance. The study reveals that while face orientation is processed unconsciously, category-based attention requires awareness.
NATURE HUMAN BEHAVIOUR
(2021)
Article
Neurosciences
M. A. Basso, S. Frey, K. A. Guerriero, B. Jarraya, S. Kastner, K. W. Koyano, D. A. Leopold, K. Murphy, C. Poirier, W. Pope, A. C. Silva, G. Tansey, L. Uhrig
Summary: Over the past decades, non-invasive imaging methods, particularly MRI, have become increasingly popular in neuroscience for studying brain structure and function. The use of MRI has not only improved surgical procedures and implants, but also enhanced diagnosis and monitoring for brain disease. Collaborating experts aim to guide researchers and veterinarians in using this imaging technology to improve the well-being and experimental outcomes for laboratory animals.
Biographical-Item
Multidisciplinary Sciences
David C. Van Essen, Sabine Kastner, Peter Bandettini
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2021)
Review
Multidisciplinary Sciences
Roel M. Willems, Marius Peelen
Summary: Context plays a significant role in cognitive processes such as perception, attention, memory, and language. It changes the neural basis of these processes, recruits additional brain regions, and is crucial for naturalistic perception and language understanding. A contextualized approach in cognitive neuroscience can lead to the discovery of new principles of the mind and brain.
Article
Biology
Miles Wischnewski, Marius Peelen
Summary: Objects can be recognized based on intrinsic features, but visibility may be limited in daily life. Object recognition is still accurate within typical scene contexts, supported by parallel processing and feedback mechanisms in the brain.
Article
Biochemistry & Molecular Biology
Surya Gayet, Marius Peelen
Summary: Humans are adept at finding objects in complex visual scenes. Current theories of attention propose that visual processing of an object of interest is enhanced by the activation of object-specific representations in the visual cortex. However, these theories fail to account for the fact that a given object will produce different retinal images depending on its location. This study used fMRI to measure brain activity in human observers as they prepared to search for objects at different distances in indoor-scene photographs. The results showed that preparatory activity incorporated contextual expectations and systematically modulated the representations of objects based on their predicted retinal images.
Article
Psychology
Lu-Chun Yeh, Marius Peelen
Summary: This study investigated how categorical and perceptual target-distractor similarity interactively affect visual search. The results showed that categorical similarity interacted with perceptual similarity, with the strongest effect observed for perceptually similar target-distractor pairs. The EEG results revealed that perceptual similarity influenced the early part of the N2pc, while categorical similarity influenced the later part. These findings provide evidence for hierarchical processing in visual search.
JOURNAL OF EXPERIMENTAL PSYCHOLOGY-HUMAN PERCEPTION AND PERFORMANCE
(2022)
Article
Neurosciences
Sushrut Thorat, Marius V. Peelen
Summary: Feature-based attention can modulate visual processing beyond spatial attention. This study used fMRI to provide evidence for spatially-global attentional modulation for human bodies. The findings suggest that responses evoked by body silhouettes are modulated by top-down attention, becoming more selective when participants search for bodies.
Article
Neurosciences
Rober Boshra, Sabine Kastner
Summary: This review explores the importance of attention in cognition and its implementation in the primate brain. The traditional top-down account suggests that control over the selection process is exerted by feedback signals from the fronto-parietal cortex, while recent studies have expanded this view by considering feature-based attention, thalamic influences, temporal network dynamics, and behavioral dynamics.
CURRENT OPINION IN NEUROBIOLOGY
(2022)
Article
Biology
Amelie Aussel, Ian C. Fiebelkorn, Sabine Kastner, Nancy J. Kopell, Benjamin Rafael Pittman-Polletta
Summary: Even during sustained attention, the processing of attended stimuli shows rhythmic fluctuations, with periods of enhanced and diminished visual processing alternating at 4 or 8 Hz. These attentional states coincide with alternating dynamical states in the LIP, FEF, and mdPul, which exhibit different activity and connectivity at theta, beta, and gamma frequencies. A computational model of FEF and LIP shows that these observed rhythms contribute to periodicity in attention and allow for peak sensitivity in visual target detection while maintaining functional flexibility when driven by experimentally-observed mdPul activity.
Article
Neurosciences
Chuyao Yan, Benedikt Ehinger, Alexis Perez-Bellido, Marius Peelen, Floris P. de Lange
Summary: This study aimed to investigate whether perceptual anticipation relies on lower-level or higher-level information. Participants were trained to associate co-occurring objects within fixed spatial arrangements and implicitly learned temporal regularities. The findings showed that humans form configuration-specific temporal expectations instead of predicting individual objects, indicating the prioritization of higher-level over lower-level information in temporal expectation.
Article
Neurosciences
Talia Brandman, Marius Peelen
Summary: Real-world scenes consist of objects and scene backgrounds, and their processing in the visual cortex interacts. Previous studies have shown that scene context can sharpen object representations, and this study demonstrates that objects can also sharpen scene representations. These effects occur at similar latencies, suggesting a common predictive processing mechanism.
Review
Psychology, Biological
Marius V. Peelen, Paul E. Downing
Summary: Multivariate pattern analysis (MVPA) has become a powerful method for analyzing functional imaging data, allowing for the testing of cognitive theories in various domains, such as perception, attention, memory, navigation, emotion, social cognition, and motor control. This review highlights the strengths of MVPA in understanding the mechanisms behind human cognition and its ability to test predictions at the item or event level, while also discussing limitations and future directions.
NATURE HUMAN BEHAVIOUR
(2023)