Scale-Aware Anchor-Free Object Detection via Curriculum Learning for Remote Sensing Images

Abstract

Accurate detection of multi-class instance objects in remote sensing images (RSIs) is a fundamental but challenging task in the field of aviation and satellite image processing, which plays a crucial role in a wide range of practical applications. Compared with the natural image-based object detection task, RSIs-based object detection still faces two main challenges: 1) The instance objects often present large variations in object size, and they are densely arranged in the given input images; 2) Complex background distributions around instance objects tend to cause boundary blurring, making it difficult to distinguish instance objects from the background, resulting in undesired feature learning interference. In this paper, to address the above challenges, we propose a novel RSI anchor-free object detection framework that consists of two key components: a cross-channel feature pyramid network (CFPN) and multiple foregroundattentive detection heads (FDHs). First, an anchor-free baseline detector with the CFPN structure is developed to extract features from different convolutional layers and incorporates these multi-scale features through parameterized cross-channel learning processes, learning the semantic relations across different scales and levels. Next, each FDH is designed to predict an attention map to enhance the features of the foreground region in RSIs. Furthermore, under this scale-aware anchor-free baseline detector structure, we design a curriculum-style optimization objective that dynamically reweights training instances during the current training epoch, enabling the detector to receive relatively easy instances that match with its current ability. Experimental results on three publicly available object detection datasets demonstrate that the proposed method outperforms existing object detection methods.Index Terms-Remote sensing images, anchor-free object detection, feature pyramid structure, foreground attention, curriculum learning.

FSOD4RSI: Few-Shot Object Detection for Remote Sensing Images via Features Aggregation and Scale Attention

“…Pang et al [21] developed a neural network model with autonomous enhancement consisting of a lightweight residual backbone as well as classifiers and detectors to enhance the network's capacity for small target detection and computational efficiency. • Regression-based approaches: Cai et al [22] designed an unanchored target detection framework for remote sensing images, including a cross-channel feature pyramid network (CFPN) and foreground attention detection heads (FDHs). CFPN can deal with a wide range of target sizes in remote sensing images, and FDHs can enhance foreground features in remote sensing images and reduce interference from complicated background information.…”

Applications of Artificial Intelligence in the Air Transport Industry: A Bibliometric and Systematic Literature Review

“…The cluster that ranks as the second largest, accounting for 17% of the studies, exhibits a significant degree of diversity. Within this cluster, numerous studies emerge from collaborations involving robotics, satellite imaging, and image processing, offering innovative techniques for intelligent navigation that can be applied to UAV control (Arrouch et al 2022;Cai et al 2021;Castagno and Atkins 2018). Another study, originating from the joint efforts of researchers from agricultural backgrounds and technologists, expands on the navigation theme.…”