지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
이용수2
Chapter 1 Introduction 1Chapter 2 Related Work 42.1 Multi-Object Tracking 42.1.1 Association methods 42.1.2 Tracking-by-detection 52.1.3 Joint and Detection and Embedding 72.2 Key-Frame Scheduling 82.3 Self-Supervised Learning 9Chapter 3 Methodology 113.1 Detector Scheduling Network 113.2 Detector Scheduler Training via Self-Supervision 143.3 Quality Measure of Tracking without Detection 143.3.1 Cardinality Measure 153.3.2 Localization Measure 163.4 Pseudo Labeling and Detector Scheduling Loss 173.5 Tracking by Detector Scheduler 18Chapter 4 Experiments 204.1 Datasets 204.2 Evaluation Metric 204.3 Implementation Details 224.4 Ablation Study 234.4.1 Detector Scheduling Network 234.4.2 Quality Measure of Tracking 244.5 Decision Scheduling Probability on Static and Dynamic Scenes 264.6 Sensitivity Analysis 274.6.1 Sensitivity Analysis of the ξ 274.6.2 Sensitivity Analysis of the θdet 284.7 Comparison Detector Scheduler with Baseline per Sequence 294.8 Comparison with the State-of-the-art Trackers 35Chapter 5 Future Works 38Chapter 6 Conclusion 40Appendix A Comparison of detection and association complexity in MOT 43Bibliography 44
0