Zobrazeno 1 - 10
of 53
pro vyhledávání: '"Babaee, Mohammadreza"'
To overcome challenges in multiple object tracking task, recent algorithms use interaction cues alongside motion and appearance features. These algorithms use graph neural networks or transformers to extract interaction features that lead to high com
Externí odkaz:
http://arxiv.org/abs/2208.03659
Multiple object tracking is a challenging problem in computer vision due to difficulty in dealing with motion prediction, occlusion handling, and object re-identification. Many recent algorithms use motion and appearance cues to overcome these challe
Externí odkaz:
http://arxiv.org/abs/2103.04147
Video representation is a key challenge in many computer vision applications such as video classification, video captioning, and video surveillance. In this paper, we propose a novel approach for video representation that captures meaningful informat
Externí odkaz:
http://arxiv.org/abs/1905.04668
Publikováno v:
In Journal of Visual Communication and Image Representation February 2023 90
In this work, we present a novel background subtraction system that uses a deep Convolutional Neural Network (CNN) to perform the segmentation. With this approach, feature engineering and parameter tuning become unnecessary since the network paramete
Externí odkaz:
http://arxiv.org/abs/1702.01731
Publikováno v:
In Pattern Recognition April 2018 76:635-649
Publikováno v:
In Neurocomputing 15 January 2016 173 Part 2:212-223
Publikováno v:
In Neurocomputing 15 January 2016 173 Part 2:245-255
Publikováno v:
In Pattern Recognition Letters 1 September 2016 80:172-178
Satellite images clustering is a challenging problem in remote sensing and machine vision, where each image content is represented by a high-dimensional feature vector. However, the feature vectors might not be appropriate to express the semantic con
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=od______1640::28448e7e44ae666ab12b64dd9a7b8fad
https://elib.dlr.de/108057/
https://elib.dlr.de/108057/