As the available wireless spectrum grows more crowded with increased usage from high bandwidth telecommunications applications, it becomes infeasible for many other users of wireless spectrum to continue operating wit...
详细信息
As the available wireless spectrum grows more crowded with increased usage from high bandwidth telecommunications applications, it becomes infeasible for many other users of wireless spectrum to continue operating with static, inflexible methods. Among these users are radar systems, which have historically been allocated large sections of bandwidth. In order to adapt and coexist with new technology in a dynamically managed environment, next generation radars must be able to adjust their spectral configuration in realtime. The research presented in this dissertation provides a framework that can be used for determining transmission constraints over both spatial direction and signal frequency. While existing research has demonstrated how to optimize radar transmitters using adjustable amplifier matching networks, such optimizations have not been able to complete quickly enough for use in real-time adaptation. To accelerate these optimizations, this dissertation presents a faster method for evaluating the performance of transmit amplifiers using a software-defined radio (SDR) and a load-pull extrapolation method using deeplearningimage completion techniques. Additionally, the accelerated optimization technique has been adapted for use with the pulse-to-pulse waveform agility paradigm of cognitive radars. Finally, the impact on Doppler detection accuracy of modifying the radar transmit chain during a coherent radar processing interval is analyzed, along with techniques for correcting the resulting distortions.
In the realm of deploying Machine learning-based Advanced Driver Assistance Systems (ML-ADAS) into real-world scenarios, adverse weather conditions pose a significant challenge. Conventional ML models trained on clear...
详细信息
ISBN:
(数字)9798331515942
ISBN:
(纸本)9798331515959
In the realm of deploying Machine learning-based Advanced Driver Assistance Systems (ML-ADAS) into real-world scenarios, adverse weather conditions pose a significant challenge. Conventional ML models trained on clear weather data falter when faced with scenarios like extreme fog or heavy rain, potentially leading to accidents and safety hazards. This paper addresses this issue by proposing a novel approach: employing a Denoising deep Neural Network as a preprocessing step to transform adverse weather images into clear weather images, thereby enhancing the robustness of ML-ADAS systems. The proposed method eliminates the need for retraining all subsequent Depp Neural Networks (DNN) in the ML-ADAS pipeline, thus saving computational resources and time. Moreover, it improves driver visualization, which is critical for safe navigation in adverse weather conditions. By leveraging the UNet architecture trained on an augmented KITTI dataset with synthetic adverse weather images, we develop the Weather UNet (WUNet) DNN to remove weather artifacts. Our study demonstrates substantial performance improvements in object detection with WUNet preprocessing under adverse weather conditions. Notably, in scenarios involving extreme fog, our proposed solution improves the mean Average Precision (mAP) score of the YOLOv8n from 4% to 70%.
作者：
Xu, LuWei, YingNortheastern Univ
Coll Informat Sci & Engn Shenyang 110819 Peoples R China Northeastern Univ
Key Lab Med Imaging Calculat Minist Educ Shenyang 110179 Peoples R China Peking Univ
Informat Technol R&D Innovat Ctr Shaoxing Peoples R China
Single image dehazing remove haze from degraded images and recover clean scenes. Prior based Methods can achieve great results on some haze images, but their performance is limited by the handcraft prior itself. Recen...
详细信息
Single image dehazing remove haze from degraded images and recover clean scenes. Prior based Methods can achieve great results on some haze images, but their performance is limited by the handcraft prior itself. Recently, many learning-based approaches has been proposed. Most of these modules rely on matching clean and haze images for training. However, such kind of real world data can be hard to get. Also the domain shift between training and testing data may affect the results. Some unsupervised methods have been proved to work on haze scenes but they still rely on handcraft priors to guide the training. In this paper, we proposed an Unsupervised Single image Dehazing method using internal learning based on the optical model of haze and other haze-like degradation images. A Pyramid deepimage strategy is used to gradually generate clean background. The entire training doesn't need any extra data or handcraft prior, and only needs the testing image itself. The proposed method is able to deal with different kinds of haze images including other haze-like degradation (like nighttimeimages and underwater images).
Liver tumor segmentation is the first step to investigate severity of liver disorder. The conventional automatic segmentation methods highly depend on hand-crafted features and priori information, which are tedious an...
详细信息
An automatic underwater object recognition system is essential to reduce the costs of underwater inspection. In this study, we propose a novel convolutional neural network architecture that is trained on underwater vi...
详细信息
An automatic underwater object recognition system is essential to reduce the costs of underwater inspection. In this study, we propose a novel convolutional neural network architecture that is trained on underwater video frames. This method is based on a modified residual neural network (ResNet) for underwater object detection. Multi-scale ResNet (M-ResNet), the modified method, improves efficiency by utilizing multi-scale operations for the accurate detection of objects of various sizes, especially small objects. The experimental results show that the proposed method yields an accuracy of 96.5% (mAP) in recognition performance. As a consequence, we propose a novel system for automatic object detection as an application for marine environments.
This paper presents a smart weeder that can be attached to the rear of an agricultural vehicle and perform weed identification and weed removal in real-time. The weed detector is installed on the side of the weeder. T...
详细信息
ISBN:
(纸本)9781728193625
This paper presents a smart weeder that can be attached to the rear of an agricultural vehicle and perform weed identification and weed removal in real-time. The weed detector is installed on the side of the weeder. The morphology-based imageprocessing technology and YoLov3-based object detection method are used to identify specific and non-specific weeds, respectively. The weeding mechanism, including a DC motor, weeding handle, chain and gear are utilized to dig out the roots of weeds. The design idea of weeding tools originates from gardening knives, which have been improved and combined with rotating mechanisms to form a weeding machine. The weed identification method has been implemented on an embedded device with high-speed graphic processing unit. The experimental results show that the weeder can operate continuously for 31 times per minute. The weed detector can accurately detect the location of weeds even when the speed of mobile platform reduce up to 90 cm/sec.
It is the age of social media and influencer culture and the importance of style and aesthetic has never been more apparent. Photo-retouching remains a critical and often, time consuming part of this process. To simpl...
详细信息
In recent years, deep convolutional neural networks have played an increasingly important role in single-image super-resolution (SR). However, with the increase of the depth and width of networks, the super-resolution...
详细信息
This study presents the OncoScan system designed to assist dermatologists in classifying skin diseases using an improved DenseNet121 model. The system was tested on the large HAM10000 dataset of over 10,000 images of ...
详细信息
ISBN:
(数字)9798350352863
ISBN:
(纸本)9798350352870
This study presents the OncoScan system designed to assist dermatologists in classifying skin diseases using an improved DenseNet121 model. The system was tested on the large HAM10000 dataset of over 10,000 images of skin lesions in seven categories. The DenseNet121 model, characterized by dense connectivity and transient layers, was compared with the InceptionV3 and EfficientNetB7 models. The results demonstrated the superiority of the DenseNet121 model, with an accuracy of 86.04% and an AUC of 96.32%. The OncoScan interface, developed using PyQt5, supports multilingual functionality including English, Russian, and Kazakh, and offers imageprocessing tools. The system's ability to handle imageprocessing and classification with ease makes it a valuable asset for dermatologic diagnostics.
暂无评论