Article
Engineering, Electrical & Electronic
Zechu Zhou, Xinyu Zhou, Zhaoyu Chen, Pinxue Guo, Qian-Yu Liu, Wenqiang Zhang
Summary: This paper proposes a novel Pixel-level Spatio-Temporal Memory (PSTM) network that efficiently organizes and utilizes temporal and spatial context information for visual object tracking. The PSTM is constructed and updated by a memory writer using a pixel-level updating strategy to maintain temporal consistency and dynamically memorize noteworthy variations. Additionally, a memory reader (PMR) is introduced to establish relationships between the object and search region and accurately estimate the object's state without a complex manual-designed mechanism.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Food Science & Technology
Isabel Gauthier, Giselle Fiestan
Summary: Food neophobia (FN) is associated with poor health and is negatively related to the ability to visually recognize images of food. This suggests that FN may contribute to the avoidance of novel foods.
FOOD QUALITY AND PREFERENCE
(2023)
Review
Neurosciences
S. P. Arun
Summary: A fundamental question for visual systems is whether image representation can be understood in terms of its components. Decomposing an image into components is challenging due to the lack of a common dictionary and the combinatorial explosion. This article describes a novel approach to evaluate compositionality at both the behavioral and neural levels, which involves creating a large number of objects by combining a small number of predefined components. The findings show that whole object representations can be predicted from components, certain components are preferred in perception, and emergent properties can be explained using compositional models.
EUROPEAN JOURNAL OF NEUROSCIENCE
(2022)
Article
Instruments & Instrumentation
Changlei Ru, Fei Wang, Tong Li, Baiming Ren, Xin Yan
Summary: An improved point cloud global descriptor, Outline-VFH, is proposed for recognition and grasping of similar workpieces. Experimental results demonstrate that the new descriptor outperforms existing descriptors in recognition and shows great potential in vision-based robot grasping applications.
REVIEW OF SCIENTIFIC INSTRUMENTS
(2021)
Article
Neurosciences
Anna Bognar, Rufin Vogels
Summary: Current models of object recognition are based on spatial representations formed from object features in retinal images. Anorthoscopic perception, which involves recognizing objects behind occlusions or through narrow slits, requires spatiotemporal integration of shape parts. Studies on monkey IT neurons showed that while the neurons signal shape identity during slit-viewing, spatiotemporal integration for whole shape perception may occur downstream to IT.
JOURNAL OF NEUROSCIENCE
(2021)
Article
Computer Science, Artificial Intelligence
Chunjie Zhang, Chao Liang, Yao Zhao
Summary: This paper proposes a novel exemplar-based, semantic guided zero-shot recognition method. By training visual and semantic sub-models, images are assigned to different classes, and an image classification model is learned by measuring visual similarity and semantic consistency. Experimental results demonstrate the effectiveness of the proposed method.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Neurosciences
Jon Walbrin, Jorge Almeida
Summary: Research indicates that distal functional connectivity is related to high-level representations for various visual categories within the occipito-temporal cortex, showing higher pattern discriminability in voxel sets strongly connected to distal brain areas. This highlights the important relationship between the complex functional organization of the occipito-temporal cortex and wider brain connectivity.
JOURNAL OF NEUROSCIENCE
(2021)
Article
Robotics
Peng Yin, Lingyun Xu, Ji Zhang, Howie Choset
Summary: The letter introduces FusionVLAD, a fusion-based network for real-time 3D place recognition, which encodes a multiview representation of sparse 3D point clouds. It consists of two parallel branches for orientation-invariant and translation-insensitive feature extraction, with a parallel fusion module to enhance the combination of region-wise feature connection between the two branches. Experiments show that FusionVLAD outperforms state-of-the-art methods in terms of accuracy and efficiency.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Neurosciences
Arielle S. Keller, Akshay Jagadeesh, Lior Bugatus, Leanne M. Williams, Kalanit Grill-Spector
Summary: This study used fMRI to investigate how attention modulates neural representations of goal-relevant stimuli in the brain. The results showed that when two objects are simultaneously viewed, the category of the attended object can be more readily decoded. After accounting for stimulus-driven variance, the correlation in residual brain activity between a cortical region and a category-selective region of VTC was higher when the preferred category was attended. This correlation was particularly strong in the right occipital, parietal, and frontal cortices. Furthermore, stronger residual correlations between a given region and VTC were associated with better visual category information decoding.
Article
Neurosciences
Mona Rosenke, Rick van Hoof, Job van den Hurk, Kalanit Grill-Spector, Rainer Goebel
Summary: This study developed and validated a functional ROI atlas of early visual and category-selective regions in human ventral and lateral occipito-temporal cortex. Cortex-based alignment showed lower between-subject variability compared to nonlinear volumetric alignment. The atlas accurately predicted the location of ventral temporal cortex ROIs and demonstrated the utility of identifying category-specific regions in healthy subjects and populations where functional localizers cannot be run.
Article
Environmental Sciences
Kento Doi, Ryuhei Hamaguchi, Yusuke Iwasawa, Masaki Onishi, Yutaka Matsuo, Ken Sakurada
Summary: We propose a robust method for object-level change detection that accurately captures scene changes in images. By designing a network to detect object-level changes and treating the change detection task as a graph matching problem, our method is more robust than previous approaches, especially for scenes with viewpoint differences. The network does not require pixel-level change annotations and detects objects that appear or disappear by extracting objects and establishing correspondences between them.
Article
Computer Science, Artificial Intelligence
Shaochuan Zhao, Tianyang Xu, Xiao-Jun Wu, Josef Kittler
Summary: This research proposes a novel visual object tracking method that improves the robustness and temporal stability of the tracker by using a cross channel correlation mechanism and a jitter metric.
INTERNATIONAL JOURNAL OF COMPUTER VISION
(2023)
Article
Computer Science, Artificial Intelligence
Tianyu Zhu, Markus Hiller, Mahsa Ehsanpour, Rongkai Ma, Tom Drummond, Ian Reid, Hamid Rezatofighi
Summary: This study presents MO3TR, an end-to-end Transformer-based online multi-object tracking framework that addresses challenges in occlusion and achieves comparable or even better results than the current state-of-the-art methods.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Neurosciences
Viola Mocz, Maryam Vaziri-Pashkam, Marvin M. Chun, Yaoda Xu
Summary: The study demonstrates a nearly orthogonal representation of object identity and nonidentity features throughout the human ventral visual processing pathway, with these nonidentity features largely untangled from the identity features early in visual processing.
JOURNAL OF NEUROSCIENCE
(2021)
Article
Robotics
Kuan Xu, Chen Wang, Chao Chen, Wei Wu, Sebastian Scherer
Summary: Object encoding and identification are crucial for robotic tasks. This letter proposes a novel object encoding method called AirCode based on a graph of key-points. It achieves robustness to viewpoint changes, scaling, occlusion, and object deformation through feature sparse encoding and object dense encoding. Experimental results show that it outperforms state-of-the-art algorithms in object identification and provides reliable semantic relocalization.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)