Can Synthetic Data Improve Object Detection Results for Remote Sensing Images?

06/09/2020

∙

Deep learning approaches require enough training samples to perform well, but it is a challenge to collect enough real training data and label them manually. In this letter, we propose the use of realistic synthetic data with a wide distribution to improve the performance of remote sensing image aircraft detection. Specifically, to increase the variability of synthetic data, we randomly set the parameters during rendering, such as the size of the instance and the class of background images. In order to make the synthetic images more realistic, we then refine the synthetic images at the pixel level using CycleGAN with real unlabeled images. We also fine-tune the model with a small amount of real data, to obtain a higher accuracy. Experiments on NWPU VHR-10, UCAS-AOD and DIOR datasets demonstrate that the proposed method can be applied for augmenting insufficient real data.

READ FULL TEXT

Can Synthetic Data Improve Object Detection Results for Remote Sensing Images?

Procedural Synthesis of Remote Sensing Images for Robust Change Detection with Neural Networks

GMM-Based Synthetic Samples for Classification of Hyperspectral Images With Limited Training Data

Detecting Olives with Synthetic or Real Data? Olive the Above

Realistically distributing object placements in synthetic training data improves the performance of vision-based object detection models

Learning class prototypes from Synthetic InSAR with Vision Transformers

Learning to Generate Synthetic Data via Compositing

Improved visible to IR image transformation using synthetic data augmentation with cycle-consistent adversarial networks

Can Synthetic Data Improve Object Detection Results for Remote Sensing Images?

Related Research

Procedural Synthesis of Remote Sensing Images for Robust Change Detection with Neural Networks

GMM-Based Synthetic Samples for Classification of Hyperspectral Images With Limited Training Data

Detecting Olives with Synthetic or Real Data? Olive the Above

Realistically distributing object placements in synthetic training data improves the performance of vision-based object detection models

Learning class prototypes from Synthetic InSAR with Vision Transformers

Learning to Generate Synthetic Data via Compositing

Improved visible to IR image transformation using synthetic data augmentation with cycle-consistent adversarial networks