Large-Scale Synthetic Urban Dataset for Aerial Scene Understanding

Abstract

The geometric extraction and semantic understanding in bird's eye view plays an important role in cyber-physical-social systems (CPSS), because it can help human or intelligent agents (IAs) to perceive larger range of environment. Moreover, due to lack of comprehensive dataset from oblique perspective, fogend deep learning algorithms for this purpose is still in blank. In this paper, we propose a novel method to generate synthetic large-scale dataset for geometric and semantic urban scene understanding from bird's eye view. There are two main steps involved, one is modeling and the other is rendering, which are processed by CityEngine and UnrealEngine4 respectively. In this way, synthetic aligned multi-model data are obtained efficiently, including spectral images, semantic labels, depth and normal maps. Specifically, terrain elevation, street graph, building style and trees distribution are all randomly generated according realistic situation, a few of handcrafted semantic labels annotated by colors spread throughout the scene, virtual cameras moved according to realistic trajectories of unmanned aerial vehicles (UAVs). For evaluation of practicability of our dataset, we manually labeled tens of aerial images downloaded from internet. And the experiment result show that, in both pure and combined mode, the dataset can improve the performance significantly.

Syn2Real Detection in the Sky: Generation and Adaptation of Synthetic Aerial Ship Images

“…Cabezas et al introduced a multi-modal synthetic dataset for this purpose [45]. Gao et al created a large-scale synthetic urban dataset specifically for unmanned aerial vehicles research and demonstrated performance improvements in comparison with real aerial images [46]. Kiefer et al applied a large-scale, high-resolution synthetic dataset to object detection tasks for unmanned aerial vehicles [47].…”

Blockchain-Aware Distributed Dynamic Monitoring: A Smart Contract for Fog-Based Drone Management in Land Surface Changes

“…It can support more edge nodes, allowing for more efficient drone deployment and node accessibility [36,37]. This includes data relay, data-to-data communication, low latency, high data transmission rate, and other technical support for other devices connected to the fog node [38].…”

Synthetic Radar Dataset Generator for Macro-Gesture Recognition

“…The latter, will first extract three one-dimensional signals out of the three feature maps. Then it will calculate the following nine features: (1) Number of zero-crossings in radial velocity, (2, 3) arguments of maximum and minimum radial velocity, (4, 5) maximum and minimum radial velocity, (6) difference between maximum and minimum angle in azimuth, (7) difference between maximum and minimum angle in elevation, (8) difference of angle in azimuth when radial velocity reached its maximum and minimum value, (9) difference of angle in elevation when radial velocity reached its maximum and minimum value. Finally, for classification we used a Multi-Layer Perceptron (MLP), with one hidden…”