The terminology "computervision"refers to the method that leverages technologies of intelligent systems such as artificial intelligence (AI) to enable computers to gather useful information from visual sour...
详细信息
The application of convolutional neural networks (CNNs) in real-time speech recognition in real-time speech improves the real-time processing capability and the ability to accurately and effectively control speech in ...
详细信息
This paper explores the impact of data augmentation via ColorJitter on the generalization capabilities of deep learning models trained on the Single Illumination Distribution dataset (SID). By systematically manipulat...
详细信息
ISBN:
(纸本)9798350386660;9798350386677
This paper explores the impact of data augmentation via ColorJitter on the generalization capabilities of deep learning models trained on the Single Illumination Distribution dataset (SID). By systematically manipulating brightness, contrast, saturation and hue, our study reveals that while enhancements in saturation and contrast yield minimal improvements, adjustments in brightness and hue significantly increase model accuracy. However, even the most effective augmentations cannot elevate the overall model performance to a practically viable level, highlighting the limitations of training with restricted lighting diversity. Moreover, our comparative analysis across convolutional neural networks (CNN) and Transformer architectures under these conditions indicates that Transformers require higher data quality and quantity for optimal performance. These findings emphasize the necessity of incorporating rich, real-world visual features into training datasets. Also, data augmentation strategies for visual Transformers require further research to fully optimize their performance in different visual environments.
A machine vision based inspection robot system is designed in this paper to address the issues of high cost and difficulty in ensuring detection accuracy in traditional manual pumped storage station crack and seepage ...
详细信息
This research paper presents a comprehensive review of key aspects related to DC microgrids, drawing insights from multiple scholarly sources. It encompasses sensitivity analysis of conversion losses, control strategi...
详细信息
Identifying a picture39;s content based on its attributes is a fundamental task in computervision known as image classification. In this article, we provide a study on the use of machine learning algorithms to cate...
详细信息
In order to solve the problem of false detection and missing detection caused by different shapes and scales of commodities in commodity image detection, a commodity image detection algorithm based on improved YOLOv5 ...
详细信息
With the development of deep learning technology, algorithms related to computervision used for autonomous driving are becoming faster and more accurate. The software architecture of autonomous driving consists of pe...
详细信息
Maintaining good water quality is fundamental for successful and sustainable pond fish farming. It directly affects the health, growth, and overall well-being of the fish population and contributes to the overall succ...
详细信息
In response to escalating energy demand and environmental concerns, renewable energy sources have emerged as crucial alternatives to conventional power generation. This paper proposes an innovative solution to address...
详细信息
暂无评论