Feature Enhancement Network for Object Detection in Optical Remote Sensing Images

Abstract

Automatic and robust object detection in remote sensing images is of vital significance in real-world applications such as land resource management and disaster rescue. However, poor performance arises when the state-of-the-art natural image detection algorithms are directly applied to remote sensing images, which largely results from the variations in object scale, aspect ratio, indistinguishable object appearances, and complex background scenario. In this paper, we propose a novel Feature Enhancement Network (FENet) for object detection in optical remote sensing images, which consists of a Dual Attention Feature Enhancement (DAFE) module and a Context Feature Enhancement (CFE) module. Specifically, the DAFE module is introduced to highlight the network to focus on the distinctive features of the objects of interest and suppress useless ones by jointly recalibrating the spatial and channel feature responses. The CFE module is designed to capture global context cues and selectively strengthen class-aware features by leveraging image-level contextual information that indicates the presence or absence of the object classes. To this end, we employ a context encoding loss to regularize the model training which promotes the object detector to understand the scene better and narrows the probable object categories in prediction. We achieve our proposed FENet by unifying DAFE and CFE into the framework of Faster R-CNN. In the experiments, we evaluate our proposed method on two large-scale remote sensing image object detection datasets including DIOR and DOTA and demonstrate its effectiveness compared with the baseline methods.

Object detection of VisDrone by stronger feature extraction FasterRCNN

“…Since the objects in remote sensing images have different scales and directions, some researchers solve these problems by enhancing feature representation. [19][20][21][37][38][39] Fu et al 37 added a feature-fusion architecture and a rotation-aware object detector to CNN to improve feature representation. An et al 19 designed rotating anchors to detect objects in any direction.…”

“…Li et al 20 predefined two different regression variables, one to predict the anchor position and one to predict the anchor angle. Cheng et al 38 proposed a context feature enhancement (CFE) module, focusing on the distinctive features of the objects of interest and suppressing useless ones. CFC-Net 21 achieved better detection performance by capturing key features.…”

Toward object alphabet augmentation for object detection in very high‐resolution satellite images

“…A recent idea proposed in Reference 35 depends on a feature attention network that tries to boost the required object features and waken undesired. Also, it tries to get the context of each feature using the Context feature enhancement network.…”

Swin-Transformer-Based YOLOv5 for Small-Object Detection in Remote Sensing Images

“…However, when these methods are used in remote sensing images, the results are always hardly satisfactory. The main reason is that remote sensing images are obtained by sensors on aerospace and aviation equipment taken from a bird’s-eye view [ 6 ]. As a result, most remote sensing images have a wide imaging range, complex background, and an imbalanced distribution of front items [ 7 ].…”