Detecting individual abandoned houses from google street view: A hierarchical deep learning approach

Keywords: Residential housing abandonment, street view, knowledge-guided deep learning, patch-based classification
Session Type: Virtual Paper Abstract
Day: Tuesday
Session Start / End Time: 3/1/2022 02:00 PM (Eastern Time (US & Canada)) - 3/1/2022 03:20 PM (Eastern Time (US & Canada))
Room: Virtual 20

Abstract

Individual-level AH detection provides essential information for fine-resolution urban studies, government decision-makers, and private sector practitioners. However, three primary conventional data sources (field data, utility data, and remote sensing data) cannot suffice to collect such fine-resolution data in a large spatial area via a cost-effective approach. To this end, Google Street View (GSV) imagery, which emerges as the mainstream open-access data source with global coverage, provides an opportunity to address this issue. Subsequently, a follow-up challenge confronting the detection of AH arises because it lacks an effective method that can discern authentic visual features from the redundant noise in GSV images. In this study, we aim to develop an effective method to detect individual-level AH from GSV imagery. Specifically, we developed a new hierarchical deep learning method to leverage global and local visual features of AH in the detection. We developed a patch-based classification method that can extract specific local features of AH. In this method, patches were generated from GSV images based on auto-detected local features, followed by being labeled as three categories: building patches, vegetation patches, and others. Individual-level AHs were detected by integrating scene classification results and patch classification results in a decision-tree model. The proposed method was applied to generate an AH map in a new site in Detroit, MI. Our study demonstrated the feasibility of GSV imagery in AH detection and showed great potential to detect AH in a large spatial extent.