Plant diseases have various impacts on the environment, encompassing ecosystems, water bodies, soils, and biodiversity. The most direct adverse effect of plant diseases on humans lies in their potential to reduce crop...
详细信息
Plant diseases have various impacts on the environment, encompassing ecosystems, water bodies, soils, and biodiversity. The most direct adverse effect of plant diseases on humans lies in their potential to reduce crop yields, posing a threat to food security. Employing deep learning models for object detection of plant diseases can effectively enhance diseases prevention and control, thereby exerting a positive influence on both ecological environment and sustainable agricultural development. However, during the detection of plant diseases, disparities in environmental conditions, locations, and data styles during data collection can result in significant variations in feature distributions between the training and testing datasets of deep learning models. In reality, the presence of such inter-domain differences leads to a notable decrease in the model's detection performance. Therefore, the primary objective of our study is to enhance the domain generalization capability of plant diseases object detection models. We employed an enhanced version of the object detector and introduced the use of Multi-Granularity Alignment (MGA) for domain adaptation, coupled with Masked image Consistency (MIC) for domain augmentation (MIC-MGA). Initially, we refined and optimized the backbone and neck sections of the object detector, and then combined it with the Multi-Granularity Alignment domain adaptation module, and finally used the Masked image Consistency module inspired by natural language processing to bolster the model's domain generalization detection capabilities. To further scrutinize the cross- domain detection robustness of MIC-MGA, we innovatively applied image style transformation techniques for data augmentation. Finally, we validated the model's performance using plant diseases data collected from various scenarios. Our approach outperformed other algorithms in all three tasks, achieving the highest mean Average Precision (mAP). Our research results affirm that the
Multispectral (MS) imaging systems have a wide range of applications for computer vision and computational photography tasks, but do not yet enjoy widespread adoption due to their prohibitive costs. Recently, advances...
详细信息
ISBN:
(纸本)9798350318920;9798350318937
Multispectral (MS) imaging systems have a wide range of applications for computer vision and computational photography tasks, but do not yet enjoy widespread adoption due to their prohibitive costs. Recently, advances in the design and fabrication of photonic metamaterials have enabled the development of MS sensors suitable for integration into consumer grade mobile devices. Augmenting existing RGB cameras and their processingalgorithms with richer spectral information has the potential to be utilized in many steps of the imageprocessing pipeline, but diverse real world datasets suitable for conducting such research are not freely available. We introduce Beyond RGB(1), a real-world dataset comprising thousands of multispectral and RGB images in diverse real world and lab conditions that is suitable for the development and evaluation of algorithms utilizing multispectral and RGB data. All the scenes in our dataset include a colorimetric reference and a measurement of the spectrum of the scene illuminant. Additionally, we demonstrate the practical use of our dataset through the introduction of a novel illuminant spectral estimation (ISE) algorithm. Our algorithm surpasses the current state-of-the-art (SoTA) by up to 58.6% on established benchmarks and sets a baseline performance on our own dataset.
This paper aims to investigate a system that uses various machine-learning algorithms to predict symptoms and deep-learning techniques for imageprocessing that leads to early disease prediction, an essential aspect o...
详细信息
In the era of rapidly expanding image data, the demand for improved image compression algorithms has grown significantly, particularly with the integration of deep learning approaches into traditional imageprocessing...
详细信息
ISBN:
(纸本)9781510673854;9781510673847
In the era of rapidly expanding image data, the demand for improved image compression algorithms has grown significantly, particularly with the integration of deep learning approaches into traditional imageprocessing tasks. However, many of the existing solutions in this domain are burdened by computational complexity, rendering them unsuitable for real-time deployment on standard devices as they often necessitate complex systems and substantial energy consumption. This work addresses the growing paradigm of edge computing for real-time applications by introducing a novel, on-edge device solution. This innovative approach aims to strike a balance between efficiency and accuracy, adhering to the practical constraints of real-world deployment. By presenting demonstrations of the proposed solution's performance on readily available devices, we provide tangible evidence of its applicability and viability in real-world scenarios. This advance contributes to the ongoing dialogue about the need for accessible and efficient image compression algorithms that can be deployed real-time applications on edge devices, bridging the gap between the demanding computational requirements of deep learning and the practical limitations of everyday hardware. As data continues to surge, solutions like this become ever more critical in ensuring effective image compression, aligning with on-edge computing within AI. This research paves the way for improved imageprocessing in real-time applications while conserving computational resources and energy consumption.
Optical Character Recognition (OCR) is a technology that integrates optical and computer technologies to convert printed text into machine-readable text. The process begins with the conversion of printed characters in...
详细信息
Biomedical image analysis has benefited tremendously from the advent of artificial intelligence. Machine learning and deep learning-based algorithms are increasingly utilized in real time to assist clinicians with mak...
详细信息
ISBN:
(纸本)9798350325744
Biomedical image analysis has benefited tremendously from the advent of artificial intelligence. Machine learning and deep learning-based algorithms are increasingly utilized in real time to assist clinicians with making crucial decisions for patients. Explainability and interpretability of these algorithms are critical for doctors and patients to develop trust in the automated decision making-process. Furthermore, designing specialized solutions based on clinical applications of these algorithms is non-trivial, due to the heterogeneity of imaging data available. Therefore, we propose an adaptive and interpretable framework for biomedical image analysis with novel applications to transcranial magnetic resonance-guided focused ultrasound thalamotomy for the treatment of essential tremor. The algorithm automatically configures itself to analyze brain lesions based on heterogeneous magnetic resonance images and subsequently predicts short-term clinical outcomes utilizing random forest and SHAP values, while ensuring interpretability for this process.
This paper seeks to design an English deep learning-based online teaching assistant system in this paper. The basic structure of the system is described in a web environment, which is composed of a system login module...
详细信息
Polarization imaging technology is widely utilized in imaging scattering media because it can capture more dimensional information compared to intensity imaging. In this study, we propose an unsupervised polarization ...
详细信息
The auxiliary recognition method based on multi-scale feature algorithms is a technology used to improve the accuracy of image recognition. Traditional image recognition algorithms often face difficulties in processin...
详细信息
Obtaining satellite images has created the need to process these images and turn them into meaningful data. Every day, new methods and algorithms are designed in the literature to meet this demand. These allow the imp...
详细信息
ISBN:
(数字)9798350388961
ISBN:
(纸本)9798350388978;9798350388961
Obtaining satellite images has created the need to process these images and turn them into meaningful data. Every day, new methods and algorithms are designed in the literature to meet this demand. These allow the improvement of parameters that evaluate the performance of imaging systems, such as the Modulation Transfer Function (MTF). Within the scope of this study, the MTF value was calculated by means of edge detection filters and mathematical operators. The algorithm was developed using the Python 3.11.1 programming language. The developed algorithm was tested for two different reference electro-optical satellite images. In this study, a literature search on the subject headings is included, and the methods and purposes of the procedures are explained.
暂无评论