6 Search Results for "Li, Hao"

Document
DOI: 10.4230/LIPIcs.GIScience.2023.7
Semi-Supervised Learning from Street-View Images and OpenStreetMap for Automatic Building Height Estimation

Authors: Hao Li, Zhendong Yuan, Gabriel Dax, Gefei Kong, Hongchao Fan, Alexander Zipf, and Martin Werner

Published in: LIPIcs, Volume 277, 12th International Conference on Geographic Information Science (GIScience 2023)

Abstract
Accurate building height estimation is key to the automatic derivation of 3D city models from emerging big geospatial data, including Volunteered Geographical Information (VGI). However, an automatic solution for large-scale building height estimation based on low-cost VGI data is currently missing. The fast development of VGI data platforms, especially OpenStreetMap (OSM) and crowdsourced street-view images (SVI), offers a stimulating opportunity to fill this research gap. In this work, we propose a semi-supervised learning (SSL) method of automatically estimating building height from Mapillary SVI and OSM data to generate low-cost and open-source 3D city modeling in LoD1. The proposed method consists of three parts: first, we propose an SSL schema with the option of setting a different ratio of "pseudo label" during the supervised regression; second, we extract multi-level morphometric features from OSM data (i.e., buildings and streets) for the purposed of inferring building height; last, we design a building floor estimation workflow with a pre-trained facade object detection network to generate "pseudo label" from SVI and assign it to the corresponding OSM building footprint. In a case study, we validate the proposed SSL method in the city of Heidelberg, Germany and evaluate the model performance against the reference data of building heights. Based on three different regression models, namely Random Forest (RF), Support Vector Machine (SVM), and Convolutional Neural Network (CNN), the SSL method leads to a clear performance boosting in estimating building heights with a Mean Absolute Error (MAE) around 2.1 meters, which is competitive to state-of-the-art approaches. The preliminary result is promising and motivates our future work in scaling up the proposed method based on low-cost VGI data, with possibilities in even regions and areas with diverse data quality and availability.
Cite as

Hao Li, Zhendong Yuan, Gabriel Dax, Gefei Kong, Hongchao Fan, Alexander Zipf, and Martin Werner. Semi-Supervised Learning from Street-View Images and OpenStreetMap for Automatic Building Height Estimation. In 12th International Conference on Geographic Information Science (GIScience 2023). Leibniz International Proceedings in Informatics (LIPIcs), Volume 277, pp. 7:1-7:15, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2023)

Copy BibTex To Clipboard

@InProceedings{li_et_al:LIPIcs.GIScience.2023.7,
  author =	{Li, Hao and Yuan, Zhendong and Dax, Gabriel and Kong, Gefei and Fan, Hongchao and Zipf, Alexander and Werner, Martin},
  title =	{{Semi-Supervised Learning from Street-View Images and OpenStreetMap for Automatic Building Height Estimation}},
  booktitle =	{12th International Conference on Geographic Information Science (GIScience 2023)},
  pages =	{7:1--7:15},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-288-4},
  ISSN =	{1868-8969},
  year =	{2023},
  volume =	{277},
  editor =	{Beecham, Roger and Long, Jed A. and Smith, Dianna and Zhao, Qunshan and Wise, Sarah},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.GIScience.2023.7},
  URN =		{urn:nbn:de:0030-drops-189028},
  doi =		{10.4230/LIPIcs.GIScience.2023.7},
  annote =	{Keywords: OpenStreetMap, Street-view Images, VGI, GeoAI, 3D city model, Facade parsing}
}
Document
DOI: 10.4230/LIPIcs.SEA.2023.2
Fast Reachability Using DAG Decomposition

Authors: Giorgos Kritikakis and Ioannis G. Tollis

Published in: LIPIcs, Volume 265, 21st International Symposium on Experimental Algorithms (SEA 2023)

Abstract
We present a fast and practical algorithm to compute the transitive closure (TC) of a directed graph. It is based on computing a reachability indexing scheme of a directed acyclic graph (DAG), G = (V, E). Given any path/chain decomposition of G we show how to compute in parameterized linear time such a reachability scheme that can answer reachability queries in constant time. The experimental results reveal that our method is significantly faster in practice than the theoretical bounds imply, indicating that path/chain decomposition algorithms can be applied to obtain fast and practical solutions to the transitive closure (TC) problem. Furthermore, we show that the number of non-transitive edges of a DAG G is ≤ width*|V| and that we can find a substantially large subset of the transitive edges of G in linear time using a path/chain decomposition. Our extensive experimental results show the interplay between these concepts in various models of DAGs.
Cite as

Giorgos Kritikakis and Ioannis G. Tollis. Fast Reachability Using DAG Decomposition. In 21st International Symposium on Experimental Algorithms (SEA 2023). Leibniz International Proceedings in Informatics (LIPIcs), Volume 265, pp. 2:1-2:17, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2023)

Copy BibTex To Clipboard

@InProceedings{kritikakis_et_al:LIPIcs.SEA.2023.2,
  author =	{Kritikakis, Giorgos and Tollis, Ioannis G.},
  title =	{{Fast Reachability Using DAG Decomposition}},
  booktitle =	{21st International Symposium on Experimental Algorithms (SEA 2023)},
  pages =	{2:1--2:17},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-279-2},
  ISSN =	{1868-8969},
  year =	{2023},
  volume =	{265},
  editor =	{Georgiadis, Loukas},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.SEA.2023.2},
  URN =		{urn:nbn:de:0030-drops-183526},
  doi =		{10.4230/LIPIcs.SEA.2023.2},
  annote =	{Keywords: graph algorithms, hierarchy, directed acyclic graphs (DAG), path/chain decomposition, transitive closure, transitive reduction, reachability, reachability indexing scheme}
}
Document
DOI: 10.4230/LITES.8.1.1
Susceptibility to Image Resolution in Face Recognition and Training Strategies to Enhance Robustness

Authors: Martin Knoche, Stefan Hörmann, and Gerhard Rigoll

Published in: LITES, Volume 8, Issue 1 (2022): Special Issue on Embedded Systems for Computer Vision. Leibniz Transactions on Embedded Systems, Volume 8, Issue 1

Abstract
Many face recognition approaches expect the input images to have similar image resolution. However, in real-world applications, the image resolution varies due to different image capture mechanisms or sources, affecting the performance of face recognition systems. This work first analyzes the image resolution susceptibility of modern face recognition. Face verification on the very popular LFW dataset drops from 99.23% accuracy to almost 55% when image dimensions of both images are reduced to arguable very poor resolution. With cross-resolution image pairs (one HR and one LR image), face verification accuracy is even worse. This characteristic is investigated more in-depth by analyzing the feature distances utilized for face verification. To increase the robustness, we propose two training strategies applied to a state-of-the-art face recognition model: 1) Training with 50% low resolution images within each batch and 2) using the cosine distance loss between high and low resolution features in a siamese network structure. Both methods significantly boost face verification accuracy for matching training and testing image resolutions. Training a network with different resolutions simultaneously instead of adding only one specific low resolution showed improvements across all resolutions and made a single model applicable to unknown resolutions. However, models trained for one particular low resolution perform better when using the exact resolution for testing. We improve the face verification accuracy from 96.86% to 97.72% on the popular LFW database with uniformly distributed image dimensions between 112 × 112 px and 5 × 5 px. Our approaches improve face verification accuracy even more from 77.56% to 87.17% for distributions focusing on lower images resolutions. Lastly, we propose specific image dimension sets focusing on high, mid, and low resolution for five well-known datasets to benchmark face verification accuracy in cross-resolution scenarios.
Cite as

Martin Knoche, Stefan Hörmann, and Gerhard Rigoll. Susceptibility to Image Resolution in Face Recognition and Training Strategies to Enhance Robustness. In LITES, Volume 8, Issue 1 (2022): Special Issue on Embedded Systems for Computer Vision. Leibniz Transactions on Embedded Systems, Volume 8, Issue 1, pp. 01:1-01:20, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2022)

Copy BibTex To Clipboard

@Article{knoche_et_al:LITES.8.1.1,
  author =	{Knoche, Martin and H\"{o}rmann, Stefan and Rigoll, Gerhard},
  title =	{{Susceptibility to Image Resolution in Face Recognition and Training Strategies to Enhance Robustness}},
  booktitle =	{LITES, Volume 8, Issue 1 (2022): Special Issue on Embedded Systems for Computer Vision},
  pages =	{01:1--01:20},
  journal =	{Leibniz Transactions on Embedded Systems},
  ISSN =	{2199-2002},
  year =	{2022},
  volume =	{8},
  number =	{1},
  editor =	{Knoche, Martin and H\"{o}rmann, Stefan and Rigoll, Gerhard},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LITES.8.1.1},
  doi =		{10.4230/LITES.8.1.1},
  annote =	{Keywords: recognition, resolution, cross, face, identification}
}
Document
DOI: 10.4230/LITES.8.1.3
HW-Flow: A Multi-Abstraction Level HW-CNN Codesign Pruning Methodology

Authors: Manoj-Rohit Vemparala, Nael Fasfous, Alexander Frickenstein, Emanuele Valpreda, Manfredi Camalleri, Qi Zhao, Christian Unger, Naveen-Shankar Nagaraja, Maurizio Martina, and Walter Stechele

Published in: LITES, Volume 8, Issue 1 (2022): Special Issue on Embedded Systems for Computer Vision. Leibniz Transactions on Embedded Systems, Volume 8, Issue 1

Abstract
Convolutional neural networks (CNNs) have produced unprecedented accuracy for many computer vision problems in the recent past. In power and compute-constrained embedded platforms, deploying modern CNNs can present many challenges. Most CNN architectures do not run in real-time due to the high number of computational operations involved during the inference phase. This emphasizes the role of CNN optimization techniques in early design space exploration. To estimate their efficacy in satisfying the target constraints, existing techniques are either hardware (HW) agnostic, pseudo-HW-aware by considering parameter and operation counts, or HW-aware through inflexible hardware-in-the-loop (HIL) setups. In this work, we introduce HW-Flow, a framework for optimizing and exploring CNN models based on three levels of hardware abstraction: Coarse, Mid and Fine. Through these levels, CNN design and optimization can be iteratively refined towards efficient execution on the target hardware platform. We present HW-Flow in the context of CNN pruning by augmenting a reinforcement learning agent with key metrics to understand the influence of its pruning actions on the inference hardware. With 2× reduction in energy and latency, we prune ResNet56, ResNet50, and DeepLabv3 with minimal accuracy degradation on the CIFAR-10, ImageNet, and CityScapes datasets, respectively.
Cite as

Manoj-Rohit Vemparala, Nael Fasfous, Alexander Frickenstein, Emanuele Valpreda, Manfredi Camalleri, Qi Zhao, Christian Unger, Naveen-Shankar Nagaraja, Maurizio Martina, and Walter Stechele. HW-Flow: A Multi-Abstraction Level HW-CNN Codesign Pruning Methodology. In LITES, Volume 8, Issue 1 (2022): Special Issue on Embedded Systems for Computer Vision. Leibniz Transactions on Embedded Systems, Volume 8, Issue 1, pp. 03:1-03:30, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2022)

Copy BibTex To Clipboard

@Article{vemparala_et_al:LITES.8.1.3,
  author =	{Vemparala, Manoj-Rohit and Fasfous, Nael and Frickenstein, Alexander and Valpreda, Emanuele and Camalleri, Manfredi and Zhao, Qi and Unger, Christian and Nagaraja, Naveen-Shankar and Martina, Maurizio and Stechele, Walter},
  title =	{{HW-Flow: A Multi-Abstraction Level HW-CNN Codesign Pruning Methodology}},
  booktitle =	{LITES, Volume 8, Issue 1 (2022): Special Issue on Embedded Systems for Computer Vision},
  pages =	{03:1--03:30},
  journal =	{Leibniz Transactions on Embedded Systems},
  ISSN =	{2199-2002},
  year =	{2022},
  volume =	{8},
  number =	{1},
  editor =	{Vemparala, Manoj-Rohit and Fasfous, Nael and Frickenstein, Alexander and Valpreda, Emanuele and Camalleri, Manfredi and Zhao, Qi and Unger, Christian and Nagaraja, Naveen-Shankar and Martina, Maurizio and Stechele, Walter},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LITES.8.1.3},
  doi =		{10.4230/LITES.8.1.3},
  annote =	{Keywords: Convolutional Neural Networks, Optimization, Hardware Modeling, Pruning}
}
Document
DOI: 10.4230/LIPIcs.DISC.2018.39
Randomized (Delta+1)-Coloring in O(log* Delta) Congested Clique Rounds

Authors: Merav Parter and Hsin-Hao Su

Published in: LIPIcs, Volume 121, 32nd International Symposium on Distributed Computing (DISC 2018)

Abstract
(Delta+1)-vertex coloring is one of the most fundamental symmetry breaking graph problems, receiving tremendous amount of attention over the last decades. We consider the congested clique model where in each round, every pair of vertices can exchange O(log n) bits of information. In a recent breakthrough, Yi-Jun Chang, Wenzheng Li, and Seth Pettie [CLP-STOC'18] presented a randomized (Delta+1)-list coloring algorithm in the LOCAL model that works in O(log^*n+Det_{deg}(log log n)) rounds, where Det_{deg}(n') is the deterministic LOCAL complexity of (deg+1)-list coloring algorithm on n'-vertex graphs. Unfortunately, the CLP algorithm uses large messages and hence cannot be efficiently implemented in the congested clique model when the maximum degree Delta is large (in particular, when Delta=omega(sqrt{n})). Merav Parter [P-ICALP'18] recently provided a randomized (Delta+1)-coloring algorithm in O(log log Delta * log^* Delta) congested clique rounds based on a careful partitioning of the input graph into almost-independent subgraphs with maximum degree sqrt{n}. In this work, we significantly improve upon this result and present a randomized (Delta+1)-coloring algorithm with O(log^* Delta) rounds, with high probability. At the heart of our algorithm is an adaptation of the CLP algorithm for coloring a subgraph with o(n) vertices and maximum degree Omega(n^{5/8}) in O(log^* Delta) rounds. The approach is built upon a combination of techniques, this includes: the graph sparsification of [Parter-ICALP'18], and a palette sampling technique adopted to the CLP framework.
Cite as

Merav Parter and Hsin-Hao Su. Randomized (Delta+1)-Coloring in O(log* Delta) Congested Clique Rounds. In 32nd International Symposium on Distributed Computing (DISC 2018). Leibniz International Proceedings in Informatics (LIPIcs), Volume 121, pp. 39:1-39:18, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2018)

Copy BibTex To Clipboard

@InProceedings{parter_et_al:LIPIcs.DISC.2018.39,
  author =	{Parter, Merav and Su, Hsin-Hao},
  title =	{{Randomized (Delta+1)-Coloring in O(log* Delta) Congested Clique Rounds}},
  booktitle =	{32nd International Symposium on Distributed Computing (DISC 2018)},
  pages =	{39:1--39:18},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-092-7},
  ISSN =	{1868-8969},
  year =	{2018},
  volume =	{121},
  editor =	{Schmid, Ulrich and Widder, Josef},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.DISC.2018.39},
  URN =		{urn:nbn:de:0030-drops-98286},
  doi =		{10.4230/LIPIcs.DISC.2018.39},
  annote =	{Keywords: Distributed Graph Algorithms, Coloring, congested clique}
}
Document
DOI: 10.4230/LIPIcs.ICALP.2018.56
A PTAS for a Class of Stochastic Dynamic Programs

Authors: Hao Fu, Jian Li, and Pan Xu

Published in: LIPIcs, Volume 107, 45th International Colloquium on Automata, Languages, and Programming (ICALP 2018)

Abstract
We develop a framework for obtaining polynomial time approximation schemes (PTAS) for a class of stochastic dynamic programs. Using our framework, we obtain the first PTAS for the following stochastic combinatorial optimization problems: 1) Probemax [Munagala, 2016]: We are given a set of n items, each item i in [n] has a value X_i which is an independent random variable with a known (discrete) distribution pi_i. We can probe a subset P subseteq [n] of items sequentially. Each time after {probing} an item i, we observe its value realization, which follows the distribution pi_i. We can adaptively probe at most m items and each item can be probed at most once. The reward is the maximum among the m realized values. Our goal is to design an adaptive probing policy such that the expected value of the reward is maximized. To the best of our knowledge, the best known approximation ratio is 1-1/e, due to Asadpour et al. [Asadpour and Nazerzadeh, 2015]. We also obtain PTAS for some generalizations and variants of the problem. 2) Committed Pandora's Box [Weitzman, 1979; Singla, 2018]: We are given a set of n boxes. For each box i in [n], the cost c_i is deterministic and the value X_i is an independent random variable with a known (discrete) distribution pi_i. Opening a box i incurs a cost of c_i. We can adaptively choose to open the boxes (and observe their values) or stop. We want to maximize the expectation of the realized value of the last opened box minus the total opening cost. 3) Stochastic Target [{I}lhan et al., 2011]: Given a predetermined target T and n items, we can adaptively insert the items into a knapsack and insert at most m items. Each item i has a value X_i which is an independent random variable with a known (discrete) distribution. Our goal is to design an adaptive policy such that the probability of the total values of all items inserted being larger than or equal to T is maximized. We provide the first bi-criteria PTAS for the problem. 4) Stochastic Blackjack Knapsack [Levin and Vainer, 2014]: We are given a knapsack of capacity C and probability distributions of n independent random variables X_i. Each item i in [n] has a size X_i and a profit p_i. We can adaptively insert the items into a knapsack, as long as the capacity constraint is not violated. We want to maximize the expected total profit of all inserted items. If the capacity constraint is violated, we lose all the profit. We provide the first bi-criteria PTAS for the problem.
Cite as

Hao Fu, Jian Li, and Pan Xu. A PTAS for a Class of Stochastic Dynamic Programs. In 45th International Colloquium on Automata, Languages, and Programming (ICALP 2018). Leibniz International Proceedings in Informatics (LIPIcs), Volume 107, pp. 56:1-56:14, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2018)

Copy BibTex To Clipboard

@InProceedings{fu_et_al:LIPIcs.ICALP.2018.56,
  author =	{Fu, Hao and Li, Jian and Xu, Pan},
  title =	{{A PTAS for a Class of Stochastic Dynamic Programs}},
  booktitle =	{45th International Colloquium on Automata, Languages, and Programming (ICALP 2018)},
  pages =	{56:1--56:14},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-076-7},
  ISSN =	{1868-8969},
  year =	{2018},
  volume =	{107},
  editor =	{Chatzigiannakis, Ioannis and Kaklamanis, Christos and Marx, D\'{a}niel and Sannella, Donald},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.ICALP.2018.56},
  URN =		{urn:nbn:de:0030-drops-90609},
  doi =		{10.4230/LIPIcs.ICALP.2018.56},
  annote =	{Keywords: stochastic optimization, dynamic program, markov decision process, block policy, approximation algorithm}
}
  • Refine by Author
  • 1 Camalleri, Manfredi
  • 1 Dax, Gabriel
  • 1 Fan, Hongchao
  • 1 Fasfous, Nael
  • 1 Frickenstein, Alexander
  • Show More...

  • Refine by Classification
  • 1 Computing methodologies → Artificial intelligence
  • 1 Computing methodologies → Neural networks
  • 1 Information systems
  • 1 Information systems → Geographic information systems
  • 1 Theory of computation → Design and analysis of algorithms
  • Show More...

  • Refine by Keyword
  • 1 3D city model
  • 1 Coloring
  • 1 Convolutional Neural Networks
  • 1 Distributed Graph Algorithms
  • 1 Facade parsing
  • Show More...

  • Refine by Type
  • 6 document

  • Refine by Publication Year
  • 2 2018
  • 2 2022
  • 2 2023

Questions / Remarks / Feedback
X

Feedback for Dagstuhl Publishing


Thanks for your feedback!

Feedback submitted

Could not send message

Please try again later or send an E-mail
© 2023 Schloss Dagstuhl - LZI GmbH Imprint Privacy Contact