The Experts below are selected from a list of 53520 Experts worldwide ranked by ideXlab platform
Daniel Cremers - One of the best experts on this subject based on the ideXlab platform.
-
gn net the gauss newton loss for multi weather relocalization
2020Co-Authors: Lukas Von Stumberg, Patrick Wenzel, Qadeer Khan, Daniel CremersAbstract:Direct SLAM methods have shown exceptional performance on odometry tasks. However, they are susceptible to dynamic lighting and weather changes while also suffering from a bad initialization on large baselines. To overcome this, we propose GN-Net: a network optimized with the novel Gauss-Newton loss for training weather invariant deep features, tailored for direct image alignment. Our network can be trained with pixel correspondences between images taken from different sequences. Experiments on both simulated and real-world datasets demonstrate that our approach is more robust against bad initialization, variations in day-time, and weather changes thereby outperforming state-of-the-art direct and indirect methods. Furthermore, we release an Evaluation Benchmark for relocalization tracking against different types of weather. Our Benchmark is available at https://vision.in.tum.de/gn-net .
-
gn net the gauss newton loss for multi weather relocalization
2019Co-Authors: Lukas Von Stumberg, Patrick Wenzel, Qadeer Khan, Daniel CremersAbstract:Direct SLAM methods have shown exceptional performance on odometry tasks. However, they are susceptible to dynamic lighting and weather changes while also suffering from a bad initialization on large baselines. To overcome this, we propose GN-Net: a network optimized with the novel Gauss-Newton loss for training weather invariant deep features, tailored for direct image alignment. Our network can be trained with pixel correspondences between images taken from different sequences. Experiments on both simulated and real-world datasets demonstrate that our approach is more robust against bad initialization, variations in day-time, and weather changes thereby outperforming state-of-the-art direct and indirect methods. Furthermore, we release an Evaluation Benchmark for relocalization tracking against different types of weather. Our Benchmark is available at this https URL.
Luc Deneire - One of the best experts on this subject based on the ideXlab platform.
-
a framework for over the air reciprocity calibration for tdd massive mimo systems
2018Co-Authors: Xiwen Jiang, Alexis Decurninge, Kalyana Gopala, Florian Kaltenberger, Maxime Guillaud, Dirk Slock, Luc DeneireAbstract:One of the biggest challenges in operating massive multiple-input multiple-output systems is the acquisition of accurate channel state information at the transmitter. To take up this challenge, time division duplex is more favorable thanks to its channel reciprocity between downlink and uplink. However, while the propagation channel over the air is reciprocal, the radio-frequency front-ends in the transceivers are not. Therefore, calibration is required to compensate the RF hardware asymmetry. Although various over-the-air calibration methods exist to address the above problem, this paper offers a unified representation of these algorithms, providing a higher level view on the calibration problem, and introduces innovations on calibration methods. We present a novel family of calibration methods, based on antenna grouping, which improves accuracy and speeds up the calibration process compared to existing methods. We then provide the Cramer–Rao bound as the performance Evaluation Benchmark and compare maximum likelihood and least squares estimators. We also differentiate between the coherent and non-coherent accumulation of calibration measurements, and point out that enabling non-coherent accumulation allows the training to be spread in time, minimizing the impact to the data service. Overall, these results have special value in allowing the design of reciprocity calibration techniques that are both accurate and resource-effective.
-
a framework for over the air reciprocity calibration for tdd massive mimo systems
2017Co-Authors: Xiwen Jiang, Alexis Decurninge, Kalyana Gopala, Florian Kaltenberger, Maxime Guillaud, Dirk Slock, Luc DeneireAbstract:One of the biggest challenges in operating massive multiple-input multiple-output systems is the acquisition of accurate channel state information at the transmitter. To take up this challenge, time division duplex is more favorable thanks to its channel reciprocity between downlink and uplink. However, while the propagation channel over the air is reciprocal, the radio-frequency front-ends in the transceivers are not. Therefore, calibration is required to compensate the RF hardware asymmetry. Although various over-the-air calibration methods exist to address the above problem, this paper offers a unified representation of these algorithms, providing a higher level view on the calibration problem, and introduces innovations on calibration methods. We present a novel family of calibration methods, based on antenna grouping, which improve accuracy and speed up the calibration process compared to existing methods. We then provide the Cram\'er-Rao bound as the performance Evaluation Benchmark and compare maximum likelihood and least squares estimators. We also differentiate between coherent and non-coherent accumulation of calibration measurements, and point out that enabling non-coherent accumulation allows the training to be spread in time, minimizing impact to the data service. Overall, these results have special value in allowing to design reciprocity calibration techniques that are both accurate and resource-effective.
Liang Chen - One of the best experts on this subject based on the ideXlab platform.
-
isles 2015 a public Evaluation Benchmark for ischemic stroke lesion segmentation from multispectral mri
2017Co-Authors: Oskar Maier, Bjoern H Menze, Janina Von Der Gablentz, Levin Hani, Mattias P Heinrich, Matthias Liebrand, Stefan Winzeck, Abdul Basit, Paul Bentley, Liang ChenAbstract:Ischemic stroke is the most common cerebrovascular disease, and its diagnosis, treatment, and study relies on non-invasive imaging. Algorithms for stroke lesion segmentation from magnetic resonance imaging (MRI) volumes are intensely researched, but the reported results are largely incomparable due to different datasets and Evaluation schemes. We approached this urgent problem of comparability with the Ischemic Stroke Lesion Segmentation (ISLES) challenge organized in conjunction with the MICCAI 2015 conference. In this paper we propose a common Evaluation framework, describe the publicly available datasets, and present the results of the two sub-challenges: Sub-Acute Stroke Lesion Segmentation (SISS) and Stroke Perfusion Estimation (SPES). A total of 16 research groups participated with a wide range of state-of-the-art automatic segmentation algorithms. A thorough analysis of the obtained data enables a critical Evaluation of the current state-of-the-art, recommendations for further developments, and the identification of remaining challenges. The segmentation of acute perfusion lesions addressed in SPES was found to be feasible. However, algorithms applied to sub-acute lesion segmentation in SISS still lack accuracy. Overall, no algorithmic characteristic of any method was found to perform superior to the others. Instead, the characteristics of stroke lesion appearances, their evolution, and the observed challenges should be studied in detail. The annotated ISLES image datasets continue to be publicly available through an online Evaluation system to serve as an ongoing Benchmarking resource (www.isles-challenge.org).
Lukas Von Stumberg - One of the best experts on this subject based on the ideXlab platform.
-
gn net the gauss newton loss for multi weather relocalization
2020Co-Authors: Lukas Von Stumberg, Patrick Wenzel, Qadeer Khan, Daniel CremersAbstract:Direct SLAM methods have shown exceptional performance on odometry tasks. However, they are susceptible to dynamic lighting and weather changes while also suffering from a bad initialization on large baselines. To overcome this, we propose GN-Net: a network optimized with the novel Gauss-Newton loss for training weather invariant deep features, tailored for direct image alignment. Our network can be trained with pixel correspondences between images taken from different sequences. Experiments on both simulated and real-world datasets demonstrate that our approach is more robust against bad initialization, variations in day-time, and weather changes thereby outperforming state-of-the-art direct and indirect methods. Furthermore, we release an Evaluation Benchmark for relocalization tracking against different types of weather. Our Benchmark is available at https://vision.in.tum.de/gn-net .
-
gn net the gauss newton loss for multi weather relocalization
2019Co-Authors: Lukas Von Stumberg, Patrick Wenzel, Qadeer Khan, Daniel CremersAbstract:Direct SLAM methods have shown exceptional performance on odometry tasks. However, they are susceptible to dynamic lighting and weather changes while also suffering from a bad initialization on large baselines. To overcome this, we propose GN-Net: a network optimized with the novel Gauss-Newton loss for training weather invariant deep features, tailored for direct image alignment. Our network can be trained with pixel correspondences between images taken from different sequences. Experiments on both simulated and real-world datasets demonstrate that our approach is more robust against bad initialization, variations in day-time, and weather changes thereby outperforming state-of-the-art direct and indirect methods. Furthermore, we release an Evaluation Benchmark for relocalization tracking against different types of weather. Our Benchmark is available at this https URL.
Xiwen Jiang - One of the best experts on this subject based on the ideXlab platform.
-
a framework for over the air reciprocity calibration for tdd massive mimo systems
2018Co-Authors: Xiwen Jiang, Alexis Decurninge, Kalyana Gopala, Florian Kaltenberger, Maxime Guillaud, Dirk Slock, Luc DeneireAbstract:One of the biggest challenges in operating massive multiple-input multiple-output systems is the acquisition of accurate channel state information at the transmitter. To take up this challenge, time division duplex is more favorable thanks to its channel reciprocity between downlink and uplink. However, while the propagation channel over the air is reciprocal, the radio-frequency front-ends in the transceivers are not. Therefore, calibration is required to compensate the RF hardware asymmetry. Although various over-the-air calibration methods exist to address the above problem, this paper offers a unified representation of these algorithms, providing a higher level view on the calibration problem, and introduces innovations on calibration methods. We present a novel family of calibration methods, based on antenna grouping, which improves accuracy and speeds up the calibration process compared to existing methods. We then provide the Cramer–Rao bound as the performance Evaluation Benchmark and compare maximum likelihood and least squares estimators. We also differentiate between the coherent and non-coherent accumulation of calibration measurements, and point out that enabling non-coherent accumulation allows the training to be spread in time, minimizing the impact to the data service. Overall, these results have special value in allowing the design of reciprocity calibration techniques that are both accurate and resource-effective.
-
a framework for over the air reciprocity calibration for tdd massive mimo systems
2017Co-Authors: Xiwen Jiang, Alexis Decurninge, Kalyana Gopala, Florian Kaltenberger, Maxime Guillaud, Dirk Slock, Luc DeneireAbstract:One of the biggest challenges in operating massive multiple-input multiple-output systems is the acquisition of accurate channel state information at the transmitter. To take up this challenge, time division duplex is more favorable thanks to its channel reciprocity between downlink and uplink. However, while the propagation channel over the air is reciprocal, the radio-frequency front-ends in the transceivers are not. Therefore, calibration is required to compensate the RF hardware asymmetry. Although various over-the-air calibration methods exist to address the above problem, this paper offers a unified representation of these algorithms, providing a higher level view on the calibration problem, and introduces innovations on calibration methods. We present a novel family of calibration methods, based on antenna grouping, which improve accuracy and speed up the calibration process compared to existing methods. We then provide the Cram\'er-Rao bound as the performance Evaluation Benchmark and compare maximum likelihood and least squares estimators. We also differentiate between coherent and non-coherent accumulation of calibration measurements, and point out that enabling non-coherent accumulation allows the training to be spread in time, minimizing impact to the data service. Overall, these results have special value in allowing to design reciprocity calibration techniques that are both accurate and resource-effective.
