In this paper we propose a novel framework for contour based object detection from cluttered environments. Given a contour model for a class of objects, it is first decomposed into fragments hierarchically. Then, we group these fragments into , where a part bundle can contain overlapping fragments. Given a new image with set of edge fragments we develop an efficient voting method using local shape similarity between part bundles and edge fragments that generates high quality candidate part configurations. We then use global shape similarity between the part configurations and the model contour to find optimal configuration. Furthermore, we show that appearance information can be used for improving detection for objects with distinctive texture when model contour does not sufficiently capture deformation of the objects.
Download full-text PDF |
Source |
---|---|
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC5328026 | PMC |
http://dx.doi.org/10.1016/j.cviu.2010.03.009 | DOI Listing |
Am J Dermatopathol
February 2025
Departments of Dermatology and Pathology, School of Medicine, Wake Forest University, Medical Center BLVD, Winston Salem, NC.
Primary vulvar carcinomas are rare and constitute a diverse group of neoplasms. These primary tumors are typically classified based on their presumed tissue of origin or histological characteristics. Among these, carcinomas of sweat gland origin are particularly significant.
View Article and Find Full Text PDFSci Rep
January 2025
Thermal and Energetic Systems Studies Laboratory (LESTE), LR99ES31, College of Engineering, University of Monastir, 5000, Monastir, Tunisia.
Using the RANS approach with the standard k-ω turbulence model, this study offers a novel investigation into the dynamic and thermal properties of turbulent impinging jet arrays. Our study examines the combined effect of the number of jets (N) and the jet-jet spacing (S) on flow mechanisms and heat transfer performance, which is unique compared to previous research that frequently focuses on the individual effects of parameters. Through the investigation of the turbulent kinetic energy, friction coefficient, velocity contours, streamlines, pressure contours, and local and mean Nusselt numbers, we provide important information about how these parameters impact flow dynamics.
View Article and Find Full Text PDFFront Endocrinol (Lausanne)
January 2025
Department of Cardiology, The Second Affiliated Hospital of Chongqing Medical University, Chongqing, China.
Background: Adrenal Vein Sampling (AVS) is the gold standard for categorizing primary aldosteronism (PA). However, catheterization of the right adrenal vein (RAV) can be technically challenging. This study aimed to investigate the validity of the right renal vertebral contour as fluoroscopic landmarks to help RAV orifice localization during AVS.
View Article and Find Full Text PDFClin Transl Radiat Oncol
March 2025
Department of Radiation Oncology, Oncology Institute of Southern Switzerland, EOC, Bellinzona, Switzerland.
Purpose: In prostate cancer patients, high radiation doses to the urethra have been associated with an increased risk of severe genitourinary toxicity following dose-escalated radiotherapy. Urethra-sparing techniques have emerged as a promising approach to reduce urinary toxicity. This international survey aims to evaluate current global practices in urethra-sparing and explore future directions for the implementation of this technique in external beam radiotherapy (EBRT) for prostate cancer.
View Article and Find Full Text PDFAesthetic Plast Surg
January 2025
Medical Research Inc., Wonju, Korea.
The pursuit of aesthetically pleasing shoulder contours, particularly those characterised by a sharp, angular definition, has gained significant traction in Southeast Asia. Traditionally, neuromodulators have been used to achieve these results by inducing muscle atrophy, particularly in the trapezius muscles. However, this approach can carry potential risks, such as compromised muscle function and spinal instability.
View Article and Find Full Text PDFEnter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!