DCL System Using Deep Learning Approaches for Land-based or Ship-based Real-time Recognition and Localization of Marine Mammals
Lead PI: Dr. Christopher Clark and Dr. Peter J. Dugan, Cornell University
The primary goals of the proposed collaborative project are to: (1) to apply and study advanced new approaches in machine learning (i.e., the convolution neural net technology, ConvNet) to bring detection and classification of marine mammals to an entirely new level of performance, (2) leverage past experiences and advanced computing to develop enhanced real-time algorithms for locating marine mammals, (3) overcome the limitations of current state-of-the-art technologies by applying, evaluating and studying a systems approach that includes the integration of the ConvNet technology with Cornell’s acoustic array processing system, Minatour, and converts these into an open and extensible real-time DCL system that can be used in a variety of environments, and (4) demonstrate the accuracy and precision of the DCL system by merging this system with a state-of-the-art towed beamforming array.
Number of Years: 3
Start Year: 2011
End Year: 2014
Partners:
- New York University
- Cornell University
- Pacific Northwest National Laboratory
- Lockheed Martin
- Hydroscience Technologies, Inc.
FY 2011 PI Report
FY 2012 PI Report
FY 2013 PI Report
FY 2014 PI Report
FY 2015 PI Report
Portable and Persistent Autonomous Real-time Marine Mammal Acoustic Monitoring
Lead PI: Dr. Christopher Clark, Cornell University
The proposed project objectives are: (1) Develop a real-time, portable, Marine Mammal Monitoring System, (2) Acquire marine acoustic data using a four-channel hydrophone array towed behind an SAIC WaveGlider, with BRP’s Pop-Up electronics integrated onto the WaveGlider, (3) Transmit selected acoustic data and associated metadata in real-time to the DMAC receiver via the SAIC high-speed satellite data link, (4)Optimize detection, classification, and localization (DCL) operations both on-board the WaveGlider and at the DMAC receiver to achieve real-time performance similar to BRP’s
current Auto-Buoy system and Pop-Up post-processing algorithms. These objectives will be achieved by demonstrating the feasibility of integrating the Pop-Up electronics and Satcom with the WaveGlider, then expanding existing DMAC capability by leveraging Auto-Buoy development, and finally implementing detection algorithms on-board the WaveGlider platform along with bearing estimates in the DMAC for real-time display of marine mammal vocalizations. Such a system would be capable of persistent, autonomous, real-time monitoring of marine mammals in areas that would otherwise not be surveyed, as it will not require a local ship for its deployment, its retrieval, or reception of data for human review.
Number of Years: 3
Start Year: 2011
End Year: 2014
Partners:
- Cornell University
- Scientific Applications International Corporation (SAIC)
FY 2011 PI Report
FY 2012 PI Report
FY 2013 PI Report
Expansion of Metadata Management, Visualization, and Data Processing Functionality of OBIS-SEAMAP
Lead PI: Dr. Patrick Halpin, Duke University
To develop the functionality required to serve as a central passive acoustics data portal, we are envisioning a framework for passive acoustic monitoring data in three tiers or levels (Figure 1). Currently OBIS-SEAMAP has focused only on the development of prototype tools for the archival and dissemination of fully processed and published marine mammal localization products. This final class of passive acoustics products would represent the most complete form or Tier 3 data in our proposed hierarchy. The core aim of this proposal is to develop a more seamless system for presenting data across multiple levels of processing and development.
Number of Years: 2
Start Year: 2011
End Year: 2014
Partners:
- Duke University
- Cornell University
- San Diego State University
- NOAA Northeast Fisheries Science Center
FY 2011 PI Report
FY 2012 PI Report
FY 2014 PI Report
Instantaneous Passive and Active Detection, Localization and Monitoring of Marine Mammals over Long Ranges
Lead PI: Dr. Purnima Ratilal, Northeastern University
The goal of this proposal is to develop approaches to enable instantaneous passive and active acoustic detection, localization and continuous monitoring of marine mammals over very wide areas spanning hundreds of kilometers or more in range. This will be accomplished using high resolution receiver array measurements of marine mammal vocalizations (passive) and instantaneous wide-area ocean acoustic waveguide remote sensing (OAWRS) and imaging (active) of marine mammals.
Number of Years: 3
Start Year: 2011
End Year: 2014
Partners:
- Northeastern University
- Massachusetts Institute of Technology
- BAE Systems
- Woods Hole Oceanographic Institution
- Defence Science and Technology Organisation
- Applied Physical Sciences
FY 2011 PI Report
FY 2012 PI Report
FY 2013 PI Report
FY 2014 PI Report
FY 2015 PI Report
Acoustic Metadata Management and Transparent Access to Networked Oceanographic Data Sets
Lead PI: Dr. Marie Roch, San Diego State University
The researchers propose to create a database system that can be used by marine mammal researchers, governments, and private industry to organize and share passive acoustic detections of marine mammals. This work is driven by the need to understand marine mammal ecology and behavior as well as the impacts of anthropogenic activities. Passive acoustic methods offer an increasingly effective technique for the monitoring of marine mammals. Advances in passive acoustic monitoring technology have produced a wealth of high sample-rate acoustic data, with multiple terabyte data sets becoming a standard practice. Analysis of these data have traditionally been accomplished through human scanning, with automated methods gaining ground in the last decade.
Number of Years: 4
Start Year: 2011
End Year: 2015
Partners:
- San Diego State University
- Scripps Institution of Oceanography
- NOAA Northeast Fisheries Science Center