UrbanDumpSight

This dataset is designed to aid in identifying and classifying improper street view dumpsites with a high degree of accuracy.ObjectivesAdherence to Criteria and Thresholds: The dataset adheres to predefined criteria and thresholds for identifying improper street view dumpsites.Classification of Dumpsite Types: It classifies dumpsite types into three categories:Construction WasteSmall Domestic WasteBulky WasteComplex and Realistic Urban Backgrounds: The dataset includes images with complex and realistic urban backgrounds to enable the model to accurately learn and distinguish between different settings, thereby minimizing false detections.Data StructureThe smallest analytical unit in the dataset is the sample point, which is labeled with three distinct identifiers for different tasks:Binary Label: Indicates the presence or absence of improper dumpsites.Waste Type Annotation: Annotates existing types of waste as Type 1, Type 2, or Type 3.Dumpsite Categorization: Categorizes the dumpsite into one of four classes:"No exist""Type 1""Type 2""Type 3"Visual RepresentationEach sample point includes four street view images taken from four angles (0, 90, 180, and 270 degrees), providing a comprehensive visual representation of the environment. These images offer crucial visual context for identifying dumpsites from multiple perspectives.MetadataA distinguishing feature of this dataset is the inclusion of extensive metadata, in addition to images and labels. This metadata encompasses:Demographic InformationSpatial Attributes related to the built environment for each sample pointThe incorporation of such detailed metadata enriches the dataset, enabling models to leverage more information about the surrounding environment to make more accurate predictions.Data Source AreaShenzhen, China