This chapter describes an experimental system for the recognition of human faces from surveillance video. In surveillance applications, the system must be robust to changes in illumination, scale, pose and expression. The system must also be able to perform detection and recognition rapidly in real time.Our system detects faces using the Viola-Jones face detector, then extracts local features to build a shape-based feature vector. The feature vector is constructed from ratios of lengths and differences in tangents of angles, so as to be robust to changes in scale and rotations in-plane and out-of-plane. Consideration was given to improving the performance and accuracy of both the detection and recognition steps.
Objectives:To evaluate the effect of maternal exposure to real-time 4-dimensional (4D) fetal ultrasound versus still 3-dimensional (3D) ultrasound on maternal-infant attachment. Methods: Three hundred pregnant women were recruited. Baseline assessments were performed at 18-20 weeks gestation and included the administration of the Maternal Antenatal Attachment Scale (MAAS) and Health Behaviour Questionnaire. Women were randomized to the 3D or 4D groups. The allocated ultrasound was performed at 28-30 weeks gestation and the Maternal Infant Attachment Questionnaire (MIA) was administered three weeks after delivery. The primary endpoint was the Quality of Attachment subscale of the MIA and comparisons between groups were adjusted for baseline MAAS scores. Results:The mean MIA Quality of Attachment subscale scores were 32.7 and 32.6 for the 3D and 4D groups, respectively (adjusted mean difference was 0.16; 95% CI: -0.67 to 0.99; p = 0.7). The mean MIA Pleasure in Interaction subscale scores were 23.4 and 23.7 for the 3D and 4D groups, respectively (adjusted mean difference was -0.33; 95% CI: -0.81 to 0.15; p = 0.2). Mean scores for the MIA Absence of Hostility Subscale were 16.0 and 16.2 for the 3D and 4D groups respectively (adjusted mean difference was -0.12 95% CI: -0.70 to 0.46; p = 0.7). Conclusions: We found that maternal exposure to real-time 4D ultrasound did not increase maternal-infant attachment. P23.08The detection of fetal motion using optical flow displacement histograms Objectives: To develop software to detect fetal motion from ultrasound sequences using a neural network and a technique called optical flow. Methods: Optical flow is the term used to describe a range of computer vision algorithms that operate under the assumption that a pixel's intensity remains constant while its position changes from x and y with dx and dy over time period dt, where dx, dy and dt represent small variations of x,y and time t. This condition is known as the brightness constancy assumption. I(x; y; t) = I(x + dx; y + dy; t + dt). Results:The following is the performance of back propagation neural network trained with a histogram input vector of size 20 at the training stage (n = 24), validation stage (n = 5) and testing stage (n = 5): mean square error = 0.08, 0.11 and 0.15 respectively; percentage error = 25,20 and 20 respectively. A reduction of input vector size from 20 to 9 required only 200 neurons (down from 250 neurons) in the hidden layer to reach the same percentage error. Conclusions: In both cases, a 20% error was reached. These results warrant further investigation of the potential of neural networks and optical flow in the detection of fetal motion from ultrasound sequences. P23.09New approaches to detect the placenta accrete using MRI and 3D Slicer T. Kanasugi, R. Oyama, G. Haba, A. Kikuchi, T. Sugiyama Obstetrics and Gynecology, Iwate Medical University, Morioka, JapanObjectives: The placenta accretes into the posterior uterine wall at third trimester, which is most difficult to make diagnosis for clinici...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.