Deep learning has achieved remarkable success across various fields, especially in imageprocessing tasks like denoising, sharpening, and contrast enhancement. However, the performance of these models heavily relies o...
详细信息
Deep learning has achieved remarkable success across various fields, especially in imageprocessing tasks like denoising, sharpening, and contrast enhancement. However, the performance of these models heavily relies on the careful selection of hyperparameters, which can be a computationally intensive and time-consuming task. Cloud-based hyperparameter search methods have gained popularity due to their ability to address the inefficiencies of single-machine training and the underutilization of computing resources. Nevertheless, these methods still encounters substantial challenges, including high computational demands, parallelism requirements, and prolonged search time. In this study, we propose HyPERTUNEFAAS, a Function as a Service (FaaS)-based hyperparameter search framework that leverages distributed computing and asynchronous processing to tackle the issues encountered in hyperparameter search. By fully exploiting the parallelism offered by serverless computing, HyPERTUNEFAAS minimizes the overhead typically associated with model training on serverless platforms. Additionally, we enhance the traditional genetic algorithm, a powerful metaheuristic method, to improve its efficiency and integrate it with the framework to enhance the efficiency of hyperparameter tuning. Experimental results demonstrate significant improvements inefficiency and cost savings with the combination of the FaaS-based hyperparameter tuning framework and the optimized genetic algorithm, making HyPERTUNEFAAS a powerful tool for optimizing image processing models and achieving superior image quality.
By combining multiple spatial influences at different levels of detail, the scale-aware spatially guided mapping (SaSGM) model can differentiate image contents in terms of scale, and models extended with SaSGM can gen...
详细信息
By combining multiple spatial influences at different levels of detail, the scale-aware spatially guided mapping (SaSGM) model can differentiate image contents in terms of scale, and models extended with SaSGM can generate more natural or diversified visual effects.
The videos or images based on deep fake video is the creation of artificial intelligence models namely deep and machine learning techniques to superimpose, replace, combine and merge images as well as clips of videos ...
详细信息
It has been proven that Logarithmic imageprocessing (LIP) models provide a suitable framework for visualizing and enhancing digital images acquired by various sources. The most visible (although simplified) result of...
详细信息
It has been proven that Logarithmic imageprocessing (LIP) models provide a suitable framework for visualizing and enhancing digital images acquired by various sources. The most visible (although simplified) result of using such a model is that LIP allows the computation of graylevel addition, subtraction and multiplication with scalars within a fixed graylevel range without the use of clipping. It is claimed that a generalized LIP framework (i.e., a parameterized family of LIP models) can be constructed on the basis of the fuzzy modelling of gray level addition as an accumulation process described by the Hamacher conorm. All the existing LIP and LIP-like models are obtained as particular cases of the proposed framework in the range corresponding to real-world digital images.
暂无评论