Sensors (Basel)
Image Processing Group, Department of Signal Theory and Communications, Universitat Politècnica de Catalunya (UPC), 08034 Barcelona, Spain.
Published: April 2022
Foreground object segmentation is a crucial first step for surveillance systems based on networks of video sensors. This problem in the context of dynamic scenes has been widely explored in the last two decades, but it still has open research questions due to challenges such as strong shadows, background clutter and illumination changes. After years of solid work based on statistical background pixel modeling, most current proposals use convolutional neural networks (CNNs) either to model the background or to make the foreground/background decision. Although these new techniques achieve outstanding results, they usually require specific training for each scene, which is unfeasible if we aim at designing software for embedded video systems and smart cameras. Our approach to the problem does not require specific context or scene training, and thus no manual labeling. We propose a network for a refinement step on top of conventional state-of-the-art background subtraction systems. By using a statistical technique to produce a rough mask, we do not need to train the network for each scene. The proposed method can take advantage of the specificity of the classic techniques, while obtaining the highly accurate segmentation that a deep learning system provides. We also show the advantage of using an adversarial network to improve the generalization ability of the network and produce more consistent results than an equivalent non-adversarial network. The results provided were obtained by training the network on a common database, without fine-tuning for specific scenes. Experiments on the unseen part of the CDNet database provided 0.82 a F-score, and 0.87 was achieved for LASIESTA databases, which is a database unrelated to the training one. On this last database, the results outperformed by 8.75% those available in the official table. The results achieved for CDNet are well above those of the methods not based on CNNs, and according to the literature, among the best for the context-unsupervised CNNs systems.
Download full-text PDF |
Source |
---|---|
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC9102692 | PMC |
http://dx.doi.org/10.3390/s22093171 | DOI Listing |
The Chern number is the core of topological photonics, which is used to describe the topological properties of photonic crystals and other optical systems to realize the functional transmission and the control of photons within materials. However, the calculation process of Chern numbers is complex and time-consuming. To address this issue, we use the deep learning accompanied with Maxwell's equations to predict the Chern number of a two-dimensional photonic crystal with a square lattice in this paper.
View Article and Find Full Text PDFOur study introduces a pioneering underwater single-pixel imaging approach that employs an orbital angular momentum (OAM) basis as a sampling scheme and a dual-attention residual U-Net generative adversarial network (DARU-GAN) as reconstruction algorithm. This method is designed to address the challenges of low sampling rates and high turbidity typically encountered in underwater environments. The integration of the OAM-basis sampling scheme and the improved reconstruction network not only enhances reconstruction quality but also ensures robust generalization capabilities, effectively restoring underwater target images even under the stringent conditions of a 3.
View Article and Find Full Text PDFSci Rep
January 2025
Division of Plastic, Craniofacial and Hand Surgery, Sidra Medicine, and Weill Cornell Medical College, C1-121, Al Gharrafa St, Ar Rayyan, Doha, Qatar.
Training a machine learning system to evaluate any type of facial deformity is impeded by the scarcity of large datasets of high-quality, ethics board-approved patient images. We have built a deep learning-based cleft lip generator called CleftGAN designed to produce an almost unlimited number of high-fidelity facsimiles of cleft lip facial images with wide variation. A transfer learning protocol testing different versions of StyleGAN as the base model was undertaken.
View Article and Find Full Text PDFPhys Med Biol
January 2025
Electrical and Computer Engineering, University of Massachusetts Lowell, Ball Hall, 1 University Ave, Lowell, Massachusetts, 01854, UNITED STATES.
Objective: X-ray photon-counting detectors (PCDs) have recently gained popularity due to their capabilities in energy discrimination power, noise suppression, and resolution refinement. The latest extremity photon-counting computed tomography (PCCT) scanner leverages these advantages for tissue characterization, material decomposition, beam hardening correction, and metal artifact reduction. However, technical challenges such as charge splitting and pulse pileup can distort the energy spectrum and compromise image quality.
View Article and Find Full Text PDFBioData Min
January 2025
School of Mathematics, Foshan University, Foshan, 528000, China.
Background: The accurate identification of molecular subtypes in digestive tract cancer (DTC) is crucial for making informed treatment decisions and selecting potential biomarkers. With the rapid advancement of artificial intelligence, various machine learning algorithms have been successfully applied in this field. However, the complexity and high dimensionality of the data features may lead to overlapping and ambiguous subtypes during clustering.
View Article and Find Full Text PDFEnter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!
© LitMetric 2025. All rights reserved.