A5045513103- Marine animal studies overview
- Wildlife Ecology and Conservation
- Underwater Acoustics Research
- Remote-Sensing Image Classification
- Species Distribution and Climate Change
- Insect and Arachnid Ecology and Behavior
- Video Surveillance and Tracking Methods
- Textile materials and evaluations
- Identification and Quantification in Food
- Animal Vocal Communication and Behavior
- Face recognition and analysis
- Ichthyology and Marine Biology
- Domain Adaptation and Few-Shot Learning
- Advanced Neural Network Applications
- Sports Dynamics and Biomechanics
- Water Quality Monitoring Technologies
- Sports Performance and Training
- Multimodal Machine Learning Applications
- Advanced Image and Video Retrieval Techniques
- Sports Analytics and Performance
- 3D Shape Modeling and Analysis
Newcastle University
2021-2023
British Antarctic Survey
2023
Abstract Researchers can investigate many aspects of animal ecology through noninvasive photo–identification. Photo–identification is becoming more efficient as matching individuals between photos increasingly automated. However, the convolutional neural network models that have facilitated this change need training images to generalize well. As a result, they often been developed for individual species meet threshold. These single‐species methods might underperform, ignore potential...
There is high demand for online fashion recommender systems that incorporate the needs of consumer's body shape. As such, we present a methodology to classify human shape from single image. This achieved through use instance segmentation and keypoint estimation models, trained only on open-source benchmarking datasets. The system capable performing in noisy environments owing robust background subtraction. proposed does not require 3D recreation as result classification based estimated...
We introduce the Northumberland Dolphin Dataset 2020 (NDD20), a challenging image dataset annotated for both coarse and fine-grained instance segmentation categorisation. This dataset, first release of NDD, was created in response to rapid expansion computer vision into conservation research production field-deployable systems suited extreme environmental conditions -- an area with few open source datasets. NDD20 contains large collection above below water images two different dolphin...
Photo-identification (photo-id) is one of the main non-invasive capture-recapture methods utilised by marine researchers for monitoring cetacean (dolphin, whale, and porpoise) populations. This method has historically been performed manually resulting in high workload cost due to vast number images collected. Recently automated aids have developed help speed-up photo-id, although they are often disjoint their processing do not utilise all available identifying information. Work presented...
Methods for cetacean research include photo-identification (photo-id) and passive acoustic monitoring (PAM) which generate thousands of images per expedition that are currently hand categorised by researchers into the individual dolphins sighted. With vast amount data obtained it is crucially important to develop a system able categorise this quickly. The Northumberland Dolphin Dataset (NDD) an on-going novel dataset project made up above below water of, spectrograms whistles from,...
Understanding the abundance of a species is first step towards understanding both its long-term sustainability and impact that we may be having upon it. Ecologists use camera traps to remotely survey for presence specific animal species. Previous studies have shown deep learning models can trained automatically detect classify animals within trap imagery with high levels confidence. However, ability train these reliant enough high-quality training data. What happens when rare or data sets...
Photo-identification (photo-id) is one of the main non-invasive capture-recapture methods utilised by marine researchers for monitoring cetacean (dolphin, whale, and porpoise) populations. This method has historically been performed manually resulting in high workload cost due to vast number images collected. Recently automated aids have developed help speed-up photo-id, although they are often disjoint their processing do not utilise all available identifying information. Work presented...
Understanding the abundance of a species is first step towards understanding both its long-term sustainability and impact that we may be having upon it. Ecologists use camera traps to remotely survey for presence specific animal species. Previous studies have shown deep learning models can trained automatically detect classify animals within trap imagery with high levels confidence. However, ability train these reliant enough high-quality training data. What happens when rare or data sets...