Note: Intro -- Preface -- Acknowledgments -- Contents -- 1 Overview of the Fish4Knowledge Project -- 1.1 Introduction -- 1.2 A Quick Tour of the Project -- 1.3 Background Information About the Studied Marine Environments -- 1.4 Project Context, Objectives and Achievements -- 1.5 Project Team -- 1.6 Conclusions -- References -- 2 User Information Needs -- 2.1 Introduction -- 2.2 Information Needs for Ecology Research on Fish Populations -- 2.3 Data Collection Techniques -- 2.4 Potential Biases -- 2.5 Uncertainty Factors Impacting the Potential Biases -- 2.6 Conclusion -- References -- 3 Supercomputing Resources -- 3.1 Introduction -- 3.2 Computational Platform -- 3.2.1 Supercomputing Platform -- 3.2.2 The Virtual Machine Cluster Platform -- 3.3 Process Execution Interface -- 3.3.1 Distributed Resource Management System -- 3.4 Summary -- References -- 4 Marine Video Data Capture and Storage -- 4.1 Introduction -- 4.2 Enhanced Video Capturing System -- 4.2.1 Better Video Server Management -- 4.2.2 Local Buffer Space -- 4.3 Massive Storage System -- 4.3.1 Assembly of Storage Drives -- 4.4 Improvement of Data Retrieval Efficiency -- 4.4.1 Universally Unique Identifier -- 4.4.2 Database Caching -- 4.5 Summary -- References -- 5 Logical Data Resource Storage -- 5.1 Introduction -- 5.2 Data Management -- 5.2.1 Design of Database -- 5.2.2 Videos Database -- 5.2.3 MetaData Database -- 5.3 Implementation -- 5.4 Future Work -- References -- 6 Software Architecture with Flexibility for the Data-Intensive Fish4Knowledge Project -- 6.1 Introduction -- 6.2 Software Design -- 6.2.1 Grand Design of Interaction -- 6.2.2 Problem Verification of the Grand Design -- 6.2.3 Practical Issues in Design Concerning the Database -- 6.2.4 Software Components Within the Fish4Knowledge Project -- 6.3 Individual Software Components and Their Relations. , 6.3.1 Fish Detection/Tracking Component -- 6.3.2 Fish Recognition -- 6.3.3 Fish Clustering -- 6.3.4 User Interface -- 6.3.5 Work-Flow -- 6.3.6 Database -- 6.3.7 Final Overview of the System -- 6.4 Software Development Process Given the Architecture -- 6.4.1 First Prototype System -- 6.4.2 Final System -- 6.4.3 Data Processing Status -- 6.5 Lessons Learned with Current Architecture -- 6.5.1 Database Definitions -- 6.5.2 Dependencies -- 6.5.3 Visualization -- 6.6 Conclusion -- References -- 7 Fish4Knowledge Database Structure, Creating and Sharing Scientific Data -- 7.1 Introduction -- 7.2 Relational Datastore Schema -- 7.3 Linked Open Data -- 7.3.1 Direct Mapping to RDF -- 7.3.2 Taiwanese Coral Reef Fish Taxonomy in SKOS -- 7.3.3 Interlinking and Alternative Representations of Direct Mapping Data -- 7.4 Current Accessibility of Data -- 7.5 Data Usage and Future Possibilities -- 7.6 Summary -- References -- 8 Intelligent Workflow Management for Fish4Knowledge Using the SWELL System -- 8.1 Introduction -- 8.2 SWELL System Design -- 8.2.1 Workflow Engine -- 8.2.2 Workflow Monitor -- 8.2.3 Workflow Evaluation -- 8.3 F4K Domain Ontologies -- 8.3.1 Goal Ontology -- 8.3.2 Video Description Ontology -- 8.3.3 Capability Ontology -- 8.4 Concluding Remarks -- References -- 9 Fish Detection -- 9.1 Introduction -- 9.2 Related Work -- 9.3 The Fish Detection Approaches -- 9.3.1 Background -- 9.3.2 A Texton-Based Kernel Density Estimation for Video Object Segmentation -- 9.4 Improving Detection Performance -- 9.4.1 Perceptual Organization Model Features -- 9.4.2 Motion Objectness -- 9.5 Performance Analysis -- 9.5.1 Fish Segmentation in Underwater Videos -- 9.5.2 Post-Processing -- 9.6 Conclusions -- References -- 10 Fish Tracking -- 10.1 Introduction -- 10.2 Literature on Fish Tracking -- 10.3 Underwater Object Tracking -- 10.4 Tracking with Covariance Modeling. , 10.4.1 Covariance-Based Tracker (COV) -- 10.4.2 Covariance-Based Particle Filter (COVPF) -- 10.5 Assessing Tracking Quality Online -- 10.6 Results -- 10.7 Conclusions -- References -- 11 Hierarchical Classification System with Reject Option for Live Fish Recognition -- 11.1 Introduction -- 11.2 Related Work -- 11.3 Feature Extraction -- 11.3.1 Image Pre-processing -- 11.3.2 Feature Extraction -- 11.4 Fish Recognition -- 11.4.1 The Balance Guaranteed Optimized Tree Method -- 11.4.2 Trajectory Voting Method -- 11.4.3 Gaussian Mixture Model For Reject Option -- 11.5 Fish Recognition Experiments -- 11.5.1 Fish Recognition Experiments Using Ground Truth Data -- 11.5.2 BGOTR Application to New Real Fish Videos -- 11.6 Conclusion -- References -- 12 Fish Behavior Analysis -- 12.1 Introduction -- 12.2 Problem Description, Definitions and Challenges -- 12.3 Literature Review on Fish Behavior Understanding -- 12.4 Proposed Methods -- 12.4.1 A Rule Based Method for Filtering Normal Fish Trajectories -- 12.4.2 Detecting Unusual Fish Trajectories Using Clustered and Labeled Data: Flat Classifier -- 12.4.3 Detecting Unusual Fish Trajectories Using Hierarchical Decomposition -- 12.4.4 Experiments and Results -- 12.5 Concluding Remarks -- References -- 13 Understanding Uncertainty Issues in the Exploration of Fish Counts -- 13.1 Introduction -- 13.2 Evaluating Uncertainty Due to Computer Vision Algorithms -- 13.3 Visualizing Uncertainty Due to Computer Vision Algorithms -- 13.3.1 Usability Issues with Computer Vision Evaluations -- 13.3.2 Preliminary User Study -- 13.3.3 Visualization Design for Non-expert Users -- 13.4 Evaluating Uncertainty Due to In-Situ System Deployment -- 13.5 Visualizing Uncertainty Due to In-Situ System Deployment -- 13.6 Uncertainty Due to both Computer Vision Algorithms and In-Situ Deployment -- 13.7 Future Work -- References. , 14 Data Groundtruthing and Crowdsourcing -- 14.1 Introduction -- 14.2 Ground Truth for Fish Detection and Tracking -- 14.2.1 Generating High Quality Annotations Using Collaborative Efforts -- 14.2.2 Experimental Results -- 14.2.3 Discussion -- 14.3 A Cluster-Based Approach to Fish Recognition -- 14.3.1 Introduction -- 14.3.2 Ground-Truth Annotation Using Automatic Clustering -- 14.3.3 Experiment -- 14.3.4 Discussion -- 14.4 Do You Need Experts in the Crowd? A Case Study in Fish Species Verification -- 14.4.1 Experiments -- 14.4.2 Results and Discussion -- 14.5 Conclusion -- References -- 15 Counting on Uncertainty: Obtaining Fish Counts from Machine Learning Decisions -- 15.1 Introduction -- 15.2 Related Work -- 15.3 Method for Estimation of Counts Based on Similarity Scores of a Classifier -- 15.3.1 Sampling Strategy -- 15.3.2 Normal Classification Process -- 15.3.3 Estimating Counts Based on Logistic Regression -- 15.3.4 Limitation in Estimations -- 15.4 Counting Fish with Logistic Regression -- 15.4.1 Experimental Datasets for Counting Fish -- 15.4.2 Results of Counting Fish with and Without Logistic Regression -- 15.5 Conclusion -- 15.6 Discussion -- References -- 16 Experiments with the Full Fish4Knowledge Dataset -- 16.1 Introduction -- 16.2 Data -- 16.3 Statistics of the Dataset -- 16.4 Discussion -- 16.5 Conclusions -- Reference -- 17 The Fish4Knowledge Virtual World Gallery -- 17.1 Introduction -- 17.2 The Fish4Knowledge Second Life Gallery---Ground Level -- 17.3 The Fish4Knowledge Second Life Gallery---Underwater Level -- 17.4 The Fish4Knowledge Virtual World Gallery in OpenSimulator -- 17.5 Conclusion -- 18 Conclusions -- 18.1 Summary of Achievements -- 18.2 Critical Assessment -- 18.3 What Lies in the Future -- 18.4 Project Publications -- 18.4.1 Fish Detection and Tracking -- 18.4.2 Fish Species Classification and Behavior Analysis. , 18.4.3 User Needs and Information Presentation -- 18.4.4 System Architecture and Overview -- 18.4.5 System Evaluation and Data Ground Truthing -- Reference -- Glossary -- Appendix A User Interface and Usage Scenario -- Appendix B Database Tables Related to F4K Workflow -- Appendix C F4K Database Schema -- Index.