Detecting anything overlooked in semantic segmentation / submitted by Robin Chan. Wuppertal, February 8, 2022
Content
- Introduction
- Theoretical Foundation
- Semantic Segmentation
- Real-World Applications
- Public Datasets and Benchmarks
- Evolution of Neural Network Architectures
- Evaluation Metrics
- Bibliography
- False Negative Detection and Reduction in Semantic Segmentation
- Cost-based Decision Rules
- Maximum Likelihood Decision Rules
- Prediction Error Meta Classification
- Controlled False-Negative Reduction of Minority Classes
- Out-of-Distribution Detection in Semantic Segmentation
- Softmax Entropy Thresholding
- Entropy Maximization and Meta Classification
- Benchmark for the Segmentation of Out-of-Distribution Objects
- Publications
- The Ethical Dilemma of Cost-based Decision Rules
- Maximum Likelihood Decision Rules for Handling Class Imbalance
- Prediction Error Meta Classification
- Controlled False Negative Reduction of Minority Classes
- Detecting Out-of-Distribution Objects via Softmax Entropy Thresholding
- Entropy Maximization and Meta Classification for Out-of-Distribution Detection
- SegmentMeIfYouCan: A Benchmark for Anomaly Segmentation
- Conclusion
- List of Figures
- Notations and Symbols
- Definitions, Propositions, Lemmas and Theorems