Machine learning, deep learning and computer vision are rapidly gaining the attention of marine scientists and conservationists due to their promising results in several fields of oceanography and marine sciences. The main advantage is getting reliable results in much less time as compared to laborious manual monitoring. The scope includes but is not limited to automatic detection and classification of fish in unconstrained underwater marine environment, coral classification, coastal morphological and morphodynamic modeling, sediment analysis, wind and wave modeling, weather prediction, ocean pollution, and resource management. However, there are challenges in the form of environmental variability, scarcity of relevant data, and noise which can compromise learning models performance. The strength and applicability of any machine learning or computer vision algorithm is judged by these challenges. In a nutshell, it is imperative that we adopt modern computer-intelligent techniques to facilitate the processing of hundreds of terabytes of data, collected at numerous marine sites globally, to properly monitor changes in the marine environment to assist conservationists and government bodies in taking necessary actions.
The motivation behind this topic is to address the key problems in automating the analysis and processing of data related to marine-related tasks, using either conventional machine learning, deep learning, or their effective combination. The acquired data is usually in the form of videos, imagery, chemical, and morphological features, and time sequences. The reliable automatic systems must be computationally efficient and robust against environmental variations.
Using effective and robust machine learning/deep learning and computer vision techniques to address the following
1. Automatic unconstrained underwater fish and coral classification.
2. Estimation of fish assemblage and biomass.
3. Suppression of static and moving background items in underwater videos to sift out objects of interest (fish and corals).
4. Sediment modeling.
5. Morphological and morphodynamical modeling.
6. Prediction models for oceanic weather, tides, and temperature fluctuation.
7. Automatic extent of pollution detection in oceans.
9. Detecting reef status and bleaching effects.
Machine learning, deep learning and computer vision are rapidly gaining the attention of marine scientists and conservationists due to their promising results in several fields of oceanography and marine sciences. The main advantage is getting reliable results in much less time as compared to laborious manual monitoring. The scope includes but is not limited to automatic detection and classification of fish in unconstrained underwater marine environment, coral classification, coastal morphological and morphodynamic modeling, sediment analysis, wind and wave modeling, weather prediction, ocean pollution, and resource management. However, there are challenges in the form of environmental variability, scarcity of relevant data, and noise which can compromise learning models performance. The strength and applicability of any machine learning or computer vision algorithm is judged by these challenges. In a nutshell, it is imperative that we adopt modern computer-intelligent techniques to facilitate the processing of hundreds of terabytes of data, collected at numerous marine sites globally, to properly monitor changes in the marine environment to assist conservationists and government bodies in taking necessary actions.
The motivation behind this topic is to address the key problems in automating the analysis and processing of data related to marine-related tasks, using either conventional machine learning, deep learning, or their effective combination. The acquired data is usually in the form of videos, imagery, chemical, and morphological features, and time sequences. The reliable automatic systems must be computationally efficient and robust against environmental variations.
Using effective and robust machine learning/deep learning and computer vision techniques to address the following
1. Automatic unconstrained underwater fish and coral classification.
2. Estimation of fish assemblage and biomass.
3. Suppression of static and moving background items in underwater videos to sift out objects of interest (fish and corals).
4. Sediment modeling.
5. Morphological and morphodynamical modeling.
6. Prediction models for oceanic weather, tides, and temperature fluctuation.
7. Automatic extent of pollution detection in oceans.
9. Detecting reef status and bleaching effects.
