Recent advancements in cabled ocean observatories have increased the quality and prevalence of underwater videos; this data enables the extraction of high-level biologically relevant information such as species' behaviours. Despite this increase in capability, most modern methods for the automatic interpretation of underwater videos focus only on the detection and counting organisms. We propose an efficient computer vision-and deep learning-based method for the detection of biological behaviours in videos. TempNet uses an encoder bridge and residual blocks to maintain model performance with a two-staged, spatial, then temporal, encoder. TempNet also presents temporal attention during spatial encoding as well as Wavelet Down-Sampling preprocessing to improve model accuracy. Although our system is designed for applications to diverse fish behaviours (i.e, is generic), we demonstrate its application to the detection of sablefish (Anoplopoma fimbria) startle events. We compare the proposed approach with a state-of-the-art end-to-end video detection method (ReMotENet) and a hybrid method previously offered exclusively for the detection of sablefish's startle events in videos from an existing dataset. Results show that our novel method comfortably outperforms the comparison baselines in multiple metrics, reaching a per-clip accuracy and precision of 80% and 0.81, respectively. This represents a relative improvement of 31% in accuracy and 27% in precision over the compared methods using this dataset. Our computational pipeline is also highly efficient, as it can process each 4-second video clip in only 38ms. Furthermore, since it does not employ features specific to sablefish startle events, our system can be easily extended to other behaviours in future works.
Large amounts of underwater imagery are constantly collected for environmental monitoring studies, as they are essential for estimating marine biodiversity and abundance. However, this collected data has variable quality due to uncontrolled environmental factors that cause blur and color casting. We attempt to address this issue by proposing two novel methods for underwater image enhancement.The first part of the thesis presents a deep learning architecture that integrates elements from classical methods to simultaneously address blurriness and color casting on underwater imagery in real time. We use two parallel architectures trained in a generative adversarial network scheme (GAN) with channel and spatial attention blocks to retrieve color, and discrete wavelength transform to preserve high-frequency components. Our experiments show that our method outperforms the state-of-the-art related works with respect to the structured similarity index metric (SSIM). Qualitative comparisons with color-checkers also show notable improvements over related works.The second part of the thesis proposes an unsupervised deep-learning approach for underwater image enhancement, which eliminates the need for reference images for training. This is an important step forward as for real (not synthetic) underwater images there is no high-quality reference available. Our method is based on a mathematical model for image dehazing. We use three networks to estimate the transmission map, the atmospheric light, and the enhanced image and propose a new compound loss function. We achieve results comparable to state-of-the-art supervised methods with respect to the SSIM while performing optimally at near real-time inference speeds.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.