Bird Eye-View to Street-View: A Survey

TL;DR

This survey reviews recent methods for synthesizing street-view images from satellite images, highlighting the need for advanced deep learning, better datasets, and improved evaluation metrics to enhance realism and diversity.

Contribution

It provides a comprehensive review of current techniques, identifies gaps in datasets and evaluation methods, and emphasizes the necessity for modern deep learning approaches in the field.

Findings

Deep learning techniques are essential for realistic image synthesis.

Current datasets are insufficient for public use.

Evaluation metrics need improvement for better assessment.

Abstract

In recent years, street view imagery has grown to become one of the most important sources of geospatial data collection and urban analytics, which facilitates generating meaningful insights and assisting in decision-making. Synthesizing a street-view image from its corresponding satellite image is a challenging task due to the significant differences in appearance and viewpoint between the two domains. In this study, we screened 20 recent research papers to provide a thorough review of the state-of-the-art of how street-view images are synthesized from their corresponding satellite counterparts. The main findings are: (i) novel deep learning techniques are required for synthesizing more realistic and accurate street-view images; (ii) more datasets need to be collected for public usage; and (iii) more specific evaluation metrics need to be investigated for evaluating the generated images appropriately. We conclude that, due to applying outdated deep learning techniques, the recent literature failed to generate detailed and diverse street-view images.