The Experts below are selected from a list of 99096 Experts worldwide ranked by ideXlab platform
Stephen Gould - One of the best experts on this subject based on the ideXlab platform.
-
multiview detection with feature Perspective Transformation
European Conference on Computer Vision, 2020Co-Authors: Yunzhong Hou, Liang Zheng, Stephen GouldAbstract:Incorporating multiple camera views for detection alleviates the impact of occlusions in crowded scenes. In a multiview detection system, we need to answer two important questions. First, how should we aggregate cues from multiple views? Second, how should we aggregate information from spatially neighboring locations? To address these questions, we introduce a novel multiview detector, MVDet. During multiview aggregation, for each location on the ground, existing methods use multiview anchor box features as representation, which potentially limits performance as pre-defined anchor boxes can be inaccurate. In contrast, via feature map Perspective Transformation, MVDet employs anchor-free representations with feature vectors directly sampled from corresponding pixels in multiple views. For spatial aggregation, different from previous methods that require design and operations outside of neural networks, MVDet takes a fully convolutional approach with large convolutional kernels on the multiview aggregated feature map. The proposed model is end-to-end learnable and achieves 88.2% MODA on Wildtrack dataset, outperforming the state-of-the-art by 14.1%. We also provide detailed analysis of MVDet on a newly introduced synthetic dataset, MultiviewX, which allows us to control the level of occlusion. Code and MultiviewX dataset are available at https://github.com/hou-yz/MVDet.
-
multiview detection with feature Perspective Transformation
arXiv: Computer Vision and Pattern Recognition, 2020Co-Authors: Yunzhong Hou, Liang Zheng, Stephen GouldAbstract:Incorporating multiple camera views for detection alleviates the impact of occlusions in crowded scenes. In a multiview system, we need to answer two important questions when dealing with ambiguities that arise from occlusions. First, how should we aggregate cues from the multiple views? Second, how should we aggregate unreliable 2D and 3D spatial information that has been tainted by occlusions? To address these questions, we propose a novel multiview detection system, MVDet. For multiview aggregation, existing methods combine anchor box features from the image plane, which potentially limits performance due to inaccurate anchor box shapes and sizes. In contrast, we take an anchor-free approach to aggregate multiview information by projecting feature maps onto the ground plane (bird's eye view). To resolve any remaining spatial ambiguity, we apply large kernel convolutions on the ground plane feature map and infer locations from detection peaks. Our entire model is end-to-end learnable and achieves 88.2% MODA on the standard Wildtrack dataset, outperforming the state-of-the-art by 14.1%. We also provide detailed analysis of MVDet on a newly introduced synthetic dataset, MultiviewX, which allows us to control the level of occlusion. Code and MultiviewX dataset are available at this https URL.
Mohamed M. Bayoumi - One of the best experts on this subject based on the ideXlab platform.
-
A dyadic wavelet affine invariant function for 2D shape recognition
IEEE Transactions on Pattern Analysis and Machine Intelligence, 2001Co-Authors: Mahmoud I. Khalil, Mohamed M. BayoumiAbstract:Dyadic wavelet transform has been used to derive an affine invariant function. First, an invariant function using two dyadic levels is derived. Then, this invariant function is used to derive another invariant function using six dyadic levels. We introduce the wavelet based conic equation. The invariant function is based on analyzing the object boundary using the dyadic wavelet transform. Experimental results on both synthetic and real data are used to demonstrate the discriminating power of the proposed invariant function. It has also been compared with some traditional methods. The stability of the proposed invariant function is examined. In addition, the stability under large Perspective Transformation is tested
-
Affine invariant object recognition using dyadic wavelet transform
2000 Canadian Conference on Electrical and Computer Engineering. Conference Proceedings. Navigating to a New Era (Cat. No.00TH8492), 2000Co-Authors: Mahmoud I. Khalil, Mohamed M. BayoumiAbstract:This paper presents an efficient technique of recognizing planar objects under affine Transformation. This technique is based on analyzing the object boundary using the dyadic wavelet transform. Two invariant functions are derived using different dyadic levels. Experimental results show that these invariant functions are more efficient than some traditional methods. In addition, these invariant functions are stable to Perspective Transformation.
Wenhsiang Tsai - One of the best experts on this subject based on the ideXlab platform.
-
Perspective Transformation invariant generalized hough transform for Perspective planar shape detection and matching
Pattern Recognition, 1997Co-Authors: Wenhsiang TsaiAbstract:Abstract In the real world, most object shapes are Perspectively transformed when imaged. How to recognize or locate such shapes in images is interesting and important. The conventional generalized Hough transform (GHT) is useful for detecting or locating translated two-dimensional (2D) planar shapes. However, it cannot be used for detecting Perspectively transformed planar shapes. A new version of the GHT, called Perspective-Transformation-invariant generalized Hough transform (PTIGHT), is proposed to remove this weakness. The PTIGHT is based on the use of a new Perspective reference table that is built up by applying both forward and inverse Perspective Transformations on a given template shape image from all viewing directions and positions. Due to the use of the point spread function to express the Perspective reference table, the required dimensionality of the Hough counting space (HCS) for the PTIGHT is reduced to 2D. After performing the PTIGHT on an input image, the peaks in the HCS whose values are larger than a threshold is picked out as the candidate locations of the Perspective shape to be detected in the input images. By performing an inverse PTIGHT on the candidates, one of the candidate locations whose corresponding shape matches best with the input shape is selected and the desired parameters of the Perspective Transformation can be obtained. Some experimental results are included to demonstrate the applicability of the proposed PTIGHT.
Wen-jing Li - One of the best experts on this subject based on the ideXlab platform.
-
hopfield neural networks for affine invariant matching
IEEE Transactions on Neural Networks, 2001Co-Authors: Wen-jing LiAbstract:The affine Transformation, which consists of rotation, translation, scaling, and shearing Transformations, can be considered as an approximation to the Perspective Transformation. Therefore, it is very important to find an effective means for establishing point correspondences under affine Transformation in many applications. In this paper, we consider the point correspondence problem as a subgraph matching problem and develop an energy formulation for affine invariant matching by a Hopfield type neural network. The fourth-order network is investigated first, then order reduction is done by incorporating the neighborhood information in the data. Thus we can use the second-order Hopfield network to perform subgraph isomorphism invariant to affine Transformation, which can be applied to an affine invariant shape recognition problem. Experimental results show the effectiveness and efficiency of the proposed method.
-
hopfield neural networks for affine invariant matching
IEEE Transactions on Neural Networks, 2001Co-Authors: Wen-jing LiAbstract:The affine Transformation, which consists of rotation, translation, scaling, and shearing Transformations, can be considered as an approximation to the Perspective Transformation. Therefore, it is very important to find an effective means for establishing point correspondences under affine Transformation in many applications. In this paper, we consider the point correspondence problem as a subgraph matching problem and develop an energy formulation for affine invariant matching by a Hopfield type neural network. The fourth-order network is investigated first, then order reduction is done by incorporating the neighborhood information in the data. Thus we can use the second-order Hopfield network to perform subgraph isomorphism invariant to affine Transformation, which can be applied to an affine invariant shape recognition problem. Experimental results show the effectiveness and efficiency of the proposed method.
Yunzhong Hou - One of the best experts on this subject based on the ideXlab platform.
-
multiview detection with feature Perspective Transformation
European Conference on Computer Vision, 2020Co-Authors: Yunzhong Hou, Liang Zheng, Stephen GouldAbstract:Incorporating multiple camera views for detection alleviates the impact of occlusions in crowded scenes. In a multiview detection system, we need to answer two important questions. First, how should we aggregate cues from multiple views? Second, how should we aggregate information from spatially neighboring locations? To address these questions, we introduce a novel multiview detector, MVDet. During multiview aggregation, for each location on the ground, existing methods use multiview anchor box features as representation, which potentially limits performance as pre-defined anchor boxes can be inaccurate. In contrast, via feature map Perspective Transformation, MVDet employs anchor-free representations with feature vectors directly sampled from corresponding pixels in multiple views. For spatial aggregation, different from previous methods that require design and operations outside of neural networks, MVDet takes a fully convolutional approach with large convolutional kernels on the multiview aggregated feature map. The proposed model is end-to-end learnable and achieves 88.2% MODA on Wildtrack dataset, outperforming the state-of-the-art by 14.1%. We also provide detailed analysis of MVDet on a newly introduced synthetic dataset, MultiviewX, which allows us to control the level of occlusion. Code and MultiviewX dataset are available at https://github.com/hou-yz/MVDet.
-
multiview detection with feature Perspective Transformation
arXiv: Computer Vision and Pattern Recognition, 2020Co-Authors: Yunzhong Hou, Liang Zheng, Stephen GouldAbstract:Incorporating multiple camera views for detection alleviates the impact of occlusions in crowded scenes. In a multiview system, we need to answer two important questions when dealing with ambiguities that arise from occlusions. First, how should we aggregate cues from the multiple views? Second, how should we aggregate unreliable 2D and 3D spatial information that has been tainted by occlusions? To address these questions, we propose a novel multiview detection system, MVDet. For multiview aggregation, existing methods combine anchor box features from the image plane, which potentially limits performance due to inaccurate anchor box shapes and sizes. In contrast, we take an anchor-free approach to aggregate multiview information by projecting feature maps onto the ground plane (bird's eye view). To resolve any remaining spatial ambiguity, we apply large kernel convolutions on the ground plane feature map and infer locations from detection peaks. Our entire model is end-to-end learnable and achieves 88.2% MODA on the standard Wildtrack dataset, outperforming the state-of-the-art by 14.1%. We also provide detailed analysis of MVDet on a newly introduced synthetic dataset, MultiviewX, which allows us to control the level of occlusion. Code and MultiviewX dataset are available at this https URL.