Higher-Order Multiple Object Tracking

Roberto D. Henschel

doi:10.51202/9783186875105

Keywords

Search publication

Bibliographic data

Copyright year: 2021
ISBN-Print: 978-3-18-387510-8
ISBN-Online: 978-3-18-687510-5
Publisher: VDI Verlag, Düsseldorf
Series: Informatik/ Kommunikation
Volume: 875
Language: German
Pages: 196
Product type: Book Titles

Titelei/Inhaltsverzeichnis Partial access Pages I - XVIII Download chapter (PDF)
1. Applications No access
2. 1. Video-based Multiple Object Tracking (MOT) No access
  2. Tracking-by-detection No access
3. 1. Errors caused by the object detector No access
  2. Challenges in discriminative features No access
  3. Combinatorial challenges No access
4. Related Work No access
5. Contributions No access
6. List of Publications No access
7. Outline No access
1. Sets, Maps, and Matrices No access
2. Probability Theory No access
3. 1. Important graph classes No access
  2. Computations on graphs No access
4. 1. Supervised learning No access
  2. Logistic regression No access
  3. Neural networks No access
5. Computational Complexity Theory No access
6. 1. Linear programming No access
  2. Binary linear programming No access
  3. Quadratic programming No access
  4. Binary quadratic programming No access
  5. Non-linear optimization No access
7. 1. Object detectors No access
  2. Appearance features No access
  3. Datasets No access
  4. MOT metrics No access
1. Introduction No access
2. 1. Related work No access
  2. Data association model for signal fusion No access
3. 1. Related work No access
  2. Frank-Wolfe Optimizer for Binary Solutions No access
4. 1. Data association model No access
  2. Experimental results No access
5. 1. Related work No access
  2. Method No access
  3. Evaluation No access
6. Conclusion No access
1. Introduction No access
2. Related Work No access
3. Problem Formulation No access
4. Constraints No access
5. Separation No access
6. Complexity No access
7. 1. Graph construction No access
  2. Pre-processing and post-processing No access
  3. Cost learning No access
  4. Implementation details on the lifted disjoint paths solver No access
  5. Experiment setup No access
  6. Benefit of long-range edges No access
  7. Ablation study on post-processing methods No access
  8. Accuracy of the fusion network No access
  9. Qualitative evaluations No access
  10. Benchmark evaluations No access
8. Conclusion No access
Conclusions No access Pages 165 - 171
Bibliography No access Pages 172 - 196