Search for: All records

Although various navigation apps are available, people who are blind or have low vision (PVIB) still face challenges to locate store entrances due to missing geospatial information in existing map services. Previously, we have developed a crowdsourcing platform to collect storefront accessibility and localization data to address the above challenges. In this paper, we have significantly improved the efficiency of data collection and user engagement in our new AI-enabled Smart DoorFront platform by designing and developing multiple important features, including a gamified credit ranking system, a volunteer contribution estimator, an AI-based pre-labeling function, and an image gallery feature. For achieving these, we integrate a specially designed deep learning model called MultiCLU into the Smart DoorFront. We also introduce an online machine learning mechanism to iteratively train the MultiCLU model, by using newly labeled storefront accessibility objects and their locations in images. Our new DoorFront platform not only significantly improves the efficiency of storefront accessibility data collection, but optimizes user experience. We have conducted interviews with six adults who are blind to better understand their daily travel challenges and their feedback indicated that the storefront accessibility data collected via the DoorFront platform would be very beneficial for them. 

The storefront accessibility can substantially impact the way people who are blind or visually impaired (BVI) travel in urban environments. Entrance localization is one of the biggest challenges to the BVI people. In addition, improperly designed staircases and obstructive store decorations can create considerable mobility challenges for BVI people, making it more difficult for them to navigate their community hence reducing their desire to travel. Unfortunately, there are few approaches to acquiring this information in advance through computational tools or services. In this paper, we propose a solution to collect large- scale accessibility data of New York City (NYC) storefronts using a crowdsourcing approach on Google Street View (GSV) panoramas. We develop a web-based crowdsourcing application, DoorFront, which enables volunteers not only to remotely label storefront accessibility data on GSV images, but also to validate the labeling result to ensure high data quality. In order to study the usability and user experience of our application, an informal beta-test is conducted and a user experience survey is designed for testing volunteers. The user feedback is very positive and indicates the high potential and usability of the proposed application. 

In this work, we present Unity Point-Cloud Interactive Core, a novel interactive point cloud rendering pipeline for the Unity Development Platform. The goal of the proposed pipeline is to expedite the development process for point cloud applications by encapsulating the rendering process as a standalone component, while maintaining flexibility through an implementable interface. The proposed pipeline allows for rendering arbitrarily large point clouds with improved performance and visual quality. First, a novel dynamic batching scheme is proposed to address the adaptive point sizing problem for level-of-detail (LOD) point cloud structures. Then, an approximate rendering algorithm is proposed to reduce overdraw by minimizing the overall number of fragment operations through an intermediate occlusion culling pass. For the purpose of analysis, the visual quality of renderings is quantified and measured by comparing against a high-quality baseline. In the experiments, the proposed pipeline maintains above 90 FPS for a 20 million point budget while achieving greater than 90% visual quality during interaction when rendering a point-cloud with more than 20 billion points. 

Social Distancing has proved a necessary measure in con- trolling the spread of Coronavirus. The CDC (Center for dis- ease control and prevention) in the United States recommends 6 feet as a safe distance between individuals. Therefore, a surveillance system capable of measuring distances between individuals can prove beneficial in limiting the spread. Video surveillance systems have already been introduced in various fields of our daily life to enhance security and protect individuals and sensitive infrastructure. In this work we present a way to use the existing video surveillance infrastructure to measure and monitor social distancing. We present a practical approach to monitor this distance using artificial intelligence and projective geometry techniques. Our approach, after initial setup, works in real-time requiring only a monocular surveillance camera feed. The proposed approach utilizes YOLO v4 neural network object detector for detecting pedestrians in the camera’s view. Projective transformation is used to localize the pedestrians on the ground. Finally, the real world distances between pedestrians is calculated and visualized with the right perspective and occlusion relations as if the distance marks are actually on the ground. All the implementation is in real-time, and is performed on python using the OPENCV libraries and the YOLO v4 neural net with pre- trained weights. Experimental results are provided to validate our approach. The code of this work will be made publicly available at GitHub upon acceptance.