Enabling Near-Zero Cost Object Detection in Remote Sensing Imagery via Progressive Self-Training

Deep learning-based object detection models rely heavily on large-scale and precise annotations for training. However, manually annotating bounding-box annotations for such data is both time-consuming and costly, especially when dealing with high-resolution satellite imagery containing densely packed small-sized objects. To alleviate the burden of manual annotation, we propose a simple yet effective approach, called progressive self-training object detection (PSTDet), to enable accurate object detection in remote sensing imagery without relying on manual annotations. Our PSTDet framework consists of two main components: initial pseudo label generation (IPLG) and progressive self-training with relabeling (PST-R). In IPLG, we leverage unsupervised image clustering, unsupervised instance detection, and geometric constraints to automatically generate high-quality bounding-box annotations for the initial training dataset. This innovative approach significantly reduces the time and expense associated with data annotation, laying a solid foundation for the subsequent progressive self-training stage. The annotations produced by IPLG serve as the training data for PST-R, which enhances the detector and pseudo labels through progressive self-training and our proposed noisy pseudo label filtering strategy (NPLFilter). Our NPLFilter purifies the quality of pseudo labels by integrating geometric constraints, prior knowledge, and category-adaptive thresholds. Experimental results demonstrate that our method achieves significant performance improvement on challenging NWPU VHR-10.v2 and DIOR datasets. Notably, our method far outperforms state-of-the-art weakly supervised methods and compares favorably with fully supervised methods.

I. Introduction

Deep learning has demonstrated impressive performance in object detection [1], [2], [3], change detection [4], [5], semantic segmentation [6], [7], and other areas. Among these, remote sensing object detection not only aids in monitoring natural resource utilization and urban development but also serves as a crucial tool in disaster monitoring and environmental protection efforts, thereby garnering widespread attention. Accurate object detection heavily relies on large-scale training data with bounding-box annotations. However, the complexity and diversity of objects in remote sensing images, combined with factors such as diverse object scales and densely distributed objects, present significant challenges for manual annotation of bounding boxes (see Fig. 1).