The complex underwater environment and the absorption and scattering of light in the water lead to color degradation and loss of detail during the underwater imaging process. To address these problems, we propose a si...
详细信息
Near-Infrared (NIR) images are widely used in a variety of low-light situations for security and safety applications. A colorised version of NIR images provide better image understanding and interpretation of features...
详细信息
ISBN:
(数字)9783031581816
ISBN:
(纸本)9783031581809;9783031581816
Near-Infrared (NIR) images are widely used in a variety of low-light situations for security and safety applications. A colorised version of NIR images provide better image understanding and interpretation of features. Because the number of NIR-RGB paired datasets is limited and often unavailable, a method to convert a given NIR image to an RGB image is highly desirable. The present work proposes an unsupervised image to image translation technique for generating colorized images (UGCI) for transforming an input NIR image to an RGB image. UGCI outperforms present NIR-RGB colorizing models and have shown approximately 57% improvement in terms of Frechet inception distance (FID) with reduced training time and less memory usage. Finally, a thorough comparative study based on different datasets is carried out to confirm superiority over leading colorization approaches in qualitative and quantitative assessments.
Structure from motion (SfM) is a fundamental task in computervision and allows recovering the 3D structure of a stationary scene from an image set. Finding robust and accurate feature matches plays a crucial role in ...
详细信息
Advanced supervised single-image dehazing models require a large number of trainable parameters and a huge amount of training data, containing a paired set of hazy images and corresponding clear images. To address thi...
详细信息
ISBN:
(纸本)9783031581731;9783031581748
Advanced supervised single-image dehazing models require a large number of trainable parameters and a huge amount of training data, containing a paired set of hazy images and corresponding clear images. To address this, knowledge distillation paves the way for training a small student network with the help of a larger teacher network. We propose an online distillation network for image dehazing in which, the teacher is an autoencoder network with feature attention blocks, and the student is a smaller autoencoder with fewer feature attention blocks. Specifically, the proposed model trains both the heavy Teacher network and the compact student network at the same time, with the student network learning the weights of the intermediate layers from the teacher network. The results of the experiments conducted on both indoor and outdoor datasets, demonstrate a significant improvement in performance compared to the state-of-the-art models on the basis of both image quality and fewer model parameters.
It is time-consuming to manually label the defects of the steel surface at the pixel-level. In this study, we aim to train a model for steel surface defect detection based on a dataset which is weakly labeled at the i...
详细信息
With the rapid development of AI hardware accelerators, applying deep learning-based algorithms to solve various low-level vision tasks on mobile devices has gradually become possible. However, two main problems still...
ISBN:
(纸本)9798350307184
With the rapid development of AI hardware accelerators, applying deep learning-based algorithms to solve various low-level vision tasks on mobile devices has gradually become possible. However, two main problems still need to be solved: task- specific algorithms make it difficult to integrate them into a single neural network architecture, and large amounts of parameters make it difficult to achieve real-time inference. To tackle these problems, we propose a novel network, SYENet, with only 6K parameters, to handle multiple low-level vision tasks on mobile devices in a real-time manner. The SYENet consists of two asymmetrical branches with simple building blocks. To effectively connect the results by asymmetrical branches, a Quadratic Connection Unit(QCU) is proposed. Furthermore, to improve performance, a new Outlier-Aware Loss is proposed to process the image. The proposed method proves its superior performance with the best PSNR as compared with other networks in real-time applications such as image Signal processing(ISP), Low-Light Enhancement(LLE), and Super-Resolution(SR) with 2K60FPS throughput on Qualcomm 8 Gen 1 mobile SoC(System-on-Chip). Particularly, for ISP task, SYENet got the highest score in MAI 2022 Learned Smartphone ISP challenge.
image segmentation is a crucial step in imageprocessing having various applications in biomedical image analysis. Segmentation of the magnetic resonance images of the brain is one such key area in biomedical image an...
详细信息
ISBN:
(纸本)9783031585340;9783031585357
image segmentation is a crucial step in imageprocessing having various applications in biomedical image analysis. Segmentation of the magnetic resonance images of the brain is one such key area in biomedical image analysis that segments various tissues in the brain and detects tumor regions. In this paper, an unsupervised rough spatial ensemble kernelized fuzzy clustering segmentation algorithm is presented for automated segmentation of magnetic resonance images of the brain. The proposed algorithm is an integration of Rough Fuzzy C Means clustering and the kernel method with a novel ensemble kernel being a combination of spherical kernel, Gaussian, and Cauchy kernels, which improves the performance of the segmentation algorithm. The proposed algorithm performs better than the existing clustering algorithms across a wide range of magnetic resonance images of the brain along with visual indications obtained from the results.
Medical image analysis based on deep learning has important research significance for accurately locating and identifying lesion targets. This article aims to address the issues of improving the detection efficiency a...
详细信息
The perforation data left by the fragments generated by the explosion of ammunition warheads on the surrounding metal plates can reflect the dispersion characteristics of the fragments, which is a commonly used ammuni...
详细信息
Drone-based imageprocessing offers valuable capabilities for surveillance, detection, and tracking in vast areas, aiding in disaster search and rescue, and monitoring artificial events like traffic jams and outdoor a...
详细信息
ISBN:
(纸本)9798350307443
Drone-based imageprocessing offers valuable capabilities for surveillance, detection, and tracking in vast areas, aiding in disaster search and rescue, and monitoring artificial events like traffic jams and outdoor activities under adversarial weather conditions. Nonetheless, this technology encounters numerous challenges, including handling variations in scales and perspectives and coping with environmental factors like sky interference and the presence of far and small objects. Besides, ensuring high visibility distance in 3D depth is crucial for safe flights in various settings, including airports, cities, and fields. However, local weather conditions can change rapidly during flights, leading to visibility issues caused by fog and clouds. Due to the cost of visibility measurement sensors, lower-cost methods using portable apparatus are desired for flight routines. Therefore, this paper proposes a camera-based visibility and weather condition estimation approach using complementary multiple Deep Learning (DL) and vision Language Models (VLM) under adversarial conditions. Experimental results show the superiority of enhanced 2D/3D captions with physical scales over SOTA VLMs.
暂无评论