Continuous long-term observation of coral reefs is essential for gathering crucial data that helps determine the state of environmental health. Coral reefs support 25% of all marine species which makes it necessary that they be monitored for good health. Traditional methods of coral reef monitoring, such as diver-based surveys and airborne observatories,...

Image processing systems gather and convey information using image sensors enabling a variety of IoT and mobile applications. IoT systems utilize image sensors to perform a multitude of various tasks that are important in all aspects of life such as detecting wildfires in forests, lifesaving medical imaging and reality headsets that use body tracking....

A substantial part of high energy consumption (> 60%) and large latency (> 90%) of conventional von-Neumann architectures can be attributed to the unavoidable data movement between the processor and main memory (DRAM). This is perhaps the major limiting factor for big data and machine learning applications whose usage is permeating into practically...

Transformers were originally designed for natural language processing but their application to other domains is rapidly gaining traction. In computer vision, convolution neural networks (CNNs) are the traditional choice of deep learning framework for image recognition tasks. Recently, Vision Transformers (ViTs) have been created a new benchmark by...

Convolution neural networks (CNNs) have made tremendous progress in various image processing fields like object recognition, segmentation, etc. This progress is primarily a result of supervised training on labeled data. The features learned by such a network are highly transferable and can be used for similar tasks. However, labeling a dataset is...

Machine learning models and neural networks in particular can be used in the field of image processing. Machine learning models are trained to better improve their output quality. During training, domain generalization is the problem of making accurate predictions on previously unseen domains, especially when these domains are very different from...

Deep learning-based generative models are an active area of research with numerous advancements in recent years. Most widely, generative models are based on convolutional neural network (CNN) architectures. In signal and image processing tasks, such as superresolution, 3D modeling, and more, implicit neural representations (INRs) can represent an image...

Advances in deep learning have resulted in state-of-the-art performance for a wide variety of computer vision tasks. The large quantity of training data and high computation resources have made convolutional neural networks (CNNs) a common backbone model for many of these tasks, including image classification, object detection, segmentation, unsupervised...

Conventional methods used for reconstructing road traffic scenes often rely on fixed sensors and cameras entrenched within road infrastructure. However, these approaches grapple with substantial limitations in terms of scope, cost-efficiency, and safeguarding individual privacy. The need to fortify road safety, strengthen urban planning initiatives,...

Synthetic aperture sonar (SAS) is an active acoustic imaging technique that coherently combines data from a moving array to form high-resolution imagery, especially of underwater environments. The moving array in SAS collects both magnitude and phase information which allows coherent integration methods to achieve resolution parallel to the sensor...