Amanda Hodgson

This paper introduces a new method to automate the detection of marine species in aerial imagery using a Machine Learning approach. Our proposed system has at its core, a convolutional neural network. We compare this trainable classifier to a handcrafted classifier based on color features, entropy and shape analysis. Experiments demonstrate that the convolutional neural network outperforms the handcrafted solution. We also introduce a negative training example-selection method for situations where the original training set consists of a collection of labeled images in which the objects of interest (positive examples) have been marked by a bounding box. We show that picking random rectangles from the background is not necessarily the best way to generate useful negative examples with respect to learning.

Monitoring and estimation of marine populations is of paramount importance for the conservation and management of sea species. Regular surveys are used to this purpose followed often by a manual counting process. This paper proposes an algorithm for automatic detection of dugongs from imagery taken in aerial surveys. Our algorithm exploits the fact that dugongs are rare in most images, therefore we determine regions of interest partially based on color rarity. This simple observation makes the system robust to changes in illumination. We also show that by applying the extended-maxima transform on red-ratio images, submerged dugongs with very fuzzy edges can be detected. Performance figures obtained here are promising in terms of degree of confidence in the detection of marine species, but more importantly our approach represents a significant step in automating this type of surveys.

This paper presents two algorithms to automate the detection of marine species in aerial imagery. An algorithm from an initial pilot study is presented in which morphology operations and colour analysis formed the basis of its working principle. A second approach is presented in which saturation channel and histogram-based shape profiling were used. We report on performance for both algorithms using datasets collected from an unmanned aerial system at an altitude of 1000ft. Early results have demonstrated recall values of 48.57% and 51.4%, and precision values of 4.01% and 4.97%.

The Murdoch University Cetacean Research Unit is conducting research into marine mammal biology and ecology within a number of broad programs across Australia's north-western bioregion. These programs include assessing: population dynamics and behaviour (abundance estimation, social structure, genetic structure and population connectivity); anthropogenic impacts (fisheries bycatch, tourism and coastal development); ecosystem interactions (foraging ecology, trophic interactions, habitat modelling, protected areas); marine mammal health (e.g. pathology, eco-toxicology, parasitology); and, new approaches to research (remote methods for abundance estimation, statistical modelling, and citizen science). The Australian Marine Mammal Centre, Chevron, the Fisheries Research and Development Corporation and WAMSI II have funded this research. MUCRU's programs are being applied to numerous taxa at specific locations from the Gascoyne to the eastern Kimberley coast. In the Gascoyne, long-term behavioural, genetic and acoustic research on Indo-Pacific bottlenose dolphins is ongoing in Shark Bay. The use of Unmanned Aerial Vehicles to conduct surveys of marine megafauna (including dugongs and humpback whales) is also occurring in Shark Bay and, potentially, further north. This year, a dedicated program will commence on Indo-Pacific humpback dolphins around the North West Cape. Aerial surveys of dugongs are being conducted from Exmouth Gulf to Barrow Island and dugongs are being satellite tagged off Onslow. Research into population dynamics and the impact of commercial fishing on common bottlenose dolphins is being carried out off the Pilbara coast, while ongoing work on the genetic connectivity of bottlenose, humpback and snubfin dolphins takes place in the Pilbara's nearshore habitats. Across the Kimberley coast, we are studying little known populations of snubfin, humpback, bottlenose and spinner dolphins, using standardised sampling regimes, such as boat-based transects, photographic identification and biopsy sampling. This will provide the first robust estimates of abundance of snubfin and humpback dolphins anywhere in Western Australia. Our research on these species will diversify this year with the introduction of passive acoustic monitoring techniques. We are working with a number of internationally renowned adjuncts associated with, for example, Duke University, University of Zurich, Aarhus University, University of Southern Denmark and North Carolina State University. Australian and Western Australian collaborating institutions include James Cook University, University of New South Wales, Australian National University, Flinders University and Curtin University.

Aerial surveys of marine mammals are routinely conducted within Australia and internationally to assess and monitor their habitat use and population status. This information is critical for conserving and managing human impacts on these species. In Australia, dugongs (dugong dugon) and humpback (Megaptera novaeangliae) whales are the most regularly surveyed using this method. The datasets produced from dugong aerial surveys have formed the basis of dugong density models that predict their critical habitat and allows for risk assessments. Aerial surveys have become increasingly used to provide baseline data and in monitoring to assess the impacts of new developments. However, the surveys are best suited for providing information over large spatial scales, and there are limitations in the temporal and spatial scale to which these surveys can detect changes in abundance and distribution of populations. The potential for Unmanned Aerial Vehicles (UAVs) to improve surveys and provide an alternative technique to assessing marine mammal populations and habitat use is currently being investigated. The first Australian UAV survey trial was conducted in Shark Bay, Western Australia in September 2010. Insitu Pacific Ltd programmed a ScanEagle UAV with a digital SLR camera payload to fly a set of transects over a 1.296 km2 area known to be frequented by dugongs. Dugongs, dolphins, turtles, whales and other marine fauna were successfully identified, and detection of marine fauna under various environmental conditions and at a range of altitudes was assessed. UAVs have the potential to increase the accuracy of detection, location and identification of species by providing a permanent, high resolution, visual records of sightings and an exact GPS position of each animal within an image. This technique eliminates issues related to availability of experienced observers and fatigue during surveys. UAVs surveys can also be conducted over a smaller spatial scale than manned surveys as the aircraft can fly lower and maintain more precise flight paths. This means that an area of interest can be surveyed at high intensity to obtain detailed marine fauna use of the area. Other advantages of UAVs include eliminating human risk and potentially reducing costs of aerial surveys.