Semi-Supervised Object Detection Framework with Object First Mixup for Remote Sensing Images

This paper proposes a Simple Semi-supervised Object Detection framework for Remote Sensing images, which is named SSOD-RS. SSOD-RS contains two parts, improved self-training and consistency regularization based on strong data augmentations with improved mixup. Firstly, as an augmentation algorithm, Object First mixup (OF-mixup) is proposed to adjust the weight of objects and the background, which expands the distribution of training samples while reducing the interference of the remote sensing complex background to the features of objects. Secondly, the strategy of training with assembling loss and fine-tuning is introduced into self-training to make the model fit the feature distribution of the true-labels after learning the features from the pseudo-labels. Experimental results demonstrate that SSOD-RS making use of unlabeled images can significantly improve the accuracy of the model.

1. Introduction

Object detection in remote sensing images, as an important part of the automatic extraction of remote sensing information’ aims to locate and identify interesting objects. In recent years, object detection models based on deep learning have been extensively developed, greatly improving prediction accuracy and speed. The core of object detection is feature extraction. The deep network can learn the features of massive labeled images, which is more capable than traditional models. However, the need for too many samples has also become a burden to the deep network.