This paper presents a comprehensive research study of the detection of U.S. traffic signs. Until now, the research in Traffic Sign Recognition systems has been centered on European traffic signs, but signs can look very different across different parts of the world, and a system that works well in Europe may indeed not work in the U.S. We go over the recent advances in traffic sign detection and discuss the differences in signs across the world. Then we present a comprehensive extension to the publicly available LISA-TS traffic sign data set, almost doubling its size, now with high-definition-quality footage. The extension is made with testing of tracking sign detection systems in mind, providing videos of traffic sign passes. We apply the Integral Channel Features and Aggregate Channel Features detection methods to U.S. traffic signs and show performance numbers outperforming all previous research on U.S. signs (while also performing similarly to the state of the art on European signs). Integral Channel Features have previously been used successfully for European signs, whereas Aggregate Channel Features have never been applied to the field of traffic signs. We take a look at the performance differences between the two methods and analyze how they perform on very distinctive signs, as well as white, rectangular signs, which tend to blend into their environment.

This paper introduces a system for estimating the attention of a driver wearing a first person view camera using salient objects to improve gaze estimation. A challenging data set of pedestrians crossing intersections has been captured using Google Glass worn by a driver. A challenge unique to first person view from cars is that the interior of the car can take up a large part of the image. The proposed system automatically filters out the dashboard of the car, along with other parts of the instrumentation. The remaining area is used as a region of interest for a pedestrian detector. Two cameras looking at the driver are used to determine the direction of the driver's gaze, by examining the eye corners and the center of the iris. This coarse gaze estimation is then linked to the detected pedestrians to determine which pedestrian the driver is focused on at any given time.

We propose a fully automatic annotation scheme that takes a raw 3D point cloud with a set of fitted CAD models as input and outputs convincing point-wise labels that can be used as cheap training data for point cloud segmentation. Compared with manual annotations, we show that our automatic labels are accurate while drastically reducing the annotation time and eliminating the need for manual intervention or dataset-specific parameters. Our labeling pipeline outputs semantic classes and soft point-wise object scores, which can either be binarized into standard one-hot-encoded labels, thresholded into weak labels with ambiguous points left unlabeled, or used directly as soft labels during training. We evaluate the label quality and segmentation performance of PointNet++ on a dataset of real industrial point clouds and Scan2CAD, a public dataset of indoor scenes. Our results indicate that reducing supervision in areas that are more difficult to label automatically is beneficial compared with the conventional approach of naively assigning a hard “best guess” label to every point.

Background: The video versus face‐to‐face preoperative anaesthetic assessment (VIDFACE) trial is a randomised, multicentre, two‐arm, superiority, assessor‐ and data analyst blinded clinical trial. The routine use of preoperative assessments is accepted as the gold standard for elective surgery, reducing both morbidity and mortality. In recent years, video‐based assessments have been investigated with a focus on patient satisfaction and enhancing efficacy. The VIDFACE trial aims to compare video‐based preoperative anaesthetic assessment with face‐to‐face assessments, with a focus on patient safety and satisfaction.Objective: To report the statistical analysis plan for the VIDFACE randomised clinical trial.Methods/Design: We will enrol 2260 adult participants undergoing elective ear‐nose‐throat or oral and maxillofacial surgery. Participants will be randomised 1:1 stratified by ‘trial site’ and ‘known or suspected malignant/haematological condition’ to either intervention (video‐based preoperative anaesthetic assessment) or control (face‐to‐face preoperative anaesthetic assessment). Two blinded statisticians will analyse the data. Before unblinding, we will write two abstracts based on the initial blinded analysis, one assuming A is intervention and B is control and one with the opposite assumption. Primary outcome is a composite outcome including five predefined serious complications. We assume a 30% relative risk reduction of the primary composite outcome. An interim analysis will take place after inclusion and follow‐up on half of the intended sample size.Conclusion: This statistical analysis plan of the VIDFACE trial will minimise analysis bias and add transparency to the statistics applied in the results of the trial.Trial Registration: ClinicalTrials.gov identifier: NCT06765538