Although existing anchor-based oriented object detection methods have achieved remarkable results, they require manual preset boxes, which introduce additional hyper-parameters and calculations. These methods often use more complex architectures for better performance, which makes them difficult to deploy on computationally constrained embedded platforms, such as satellites and unmanned aerial vehicles. We aim to design a high-performance algorithm that is simple, fast, and easy to deploy for aerial image detection. In this article, we propose a one-stage anchor-free rotated object detector, FCOSR, that can be deployed on most platforms and uses our well-defined label assignment strategy for the features of the aerial image objects. We use the ellipse center sampling method to define a suitable sampling region for an oriented bounding box (OBB). The fuzzy sample assignment strategy provides reasonable labels for overlapping objects. To solve the problem of insufficient sampling, we designed a multi-level sampling module. These strategies allocate more appropriate labels to training samples. Our algorithm achieves an mean average precision (mAP) of 79.25, 75.41, and 90.13 on the DOTA-v1.0, DOTA-v1.5, and HRSC2016 datasets, respectively. FCOSR demonstrates a performance superior to that of other methods in single-scale evaluation, where the small model achieves an mAP of 74.05 at a speed of 23.7 FPS on an RTX 2080-Ti GPU. When we convert the lightweight FCOSR model to the TensorRT format, it achieves an mAP of 73.93 on DOTA-v1.0 at a speed of 17.76 FPS on a Jetson AGX Xavier device with a single scale.

Ship detection in synthetic aperture radar (SAR) images has important application value. Sea clutter, complex scenes, a large size change in ships, and the arbitrary directionality of ships make ship detection challenging. With the development of deep learning, many deep learning algorithms have been applied to SAR images. These algorithms need a lot of labeled data for training. It is time-consuming to label SAR data, and the unlabeled data are easy to obtain. It is necessary to use the unlabeled data effectively to improve the performance of the algorithm. In this study, a semisupervised SAR ship detection network, named the semisupervised consistency learning adversarial network (SCLANet), is presented. SCLANet is a two-stage detection network. The local features around the ship can be extracted by the SCLANet, and the features generated from unlabeled data become closer to those generated from labeled data by using adversarial learning. There are two consistency learning modules in SCLANet: noise robustness consistency learning and output encoding consistency learning. Noise robustness consistency learning can increase the robustness of the SCLANet. Maintaining consistency between the noisy results and the original results can train the unlabeled data. In output encoding consistency learning, outputs are mapped to a picture that is fed into an encoder to obtain the intermediate representation embedding. Another embedding is a layer in the main network of the SCLANet. Reducing the error between two embeddings can train the SCLANet with unlabeled data. Two types of consistency learning can be used as pretext tasks for semisupervised learning. Experiments were conducted on two SAR ship datasets. Compared with other algorithms, the SCLANet achieved the highest detection accuracy, indicating that it is more advantageous to use in ship detection.

In aerial image scenes, the objects have properties of arbitrary orientation, large-scale range, and dense distribution. Thus, the object detector uses oriented bounding box (OBB) to locate objects, which is more complex and challenging than horizontal bounding box (HBB) detector. Mainstream OBB detectors mostly use one-to-many label assignment strategy to predict multiple bounding boxes for the same object, and filter out repeat predictions by non-maximum suppression (NMS). NMS ranks with confidence and drops the detection box with IoU higher than the threshold, which is easy to get the local optimum result. The clustered synthesis method gets more accurate results than the original NMS, but applying it to the OBB detector leads to border shift, which arises from the angular discontinuity problem. Therefore, we use Gaussian OBB (G-OBB) to deal with the angular discontinuity and thus eliminate the offset generated by direct synthesis. G-OBB is not an easy to understand and describe representation. For this reason, we analyze the properties of G-OBB, and design a decoding method to convert a G-OBB to a rotated rectangular box, further discussing its conditions. Based on the decoding method, we propose a Gaussian synthesis algorithm (GauS), which transforms the OBB into Gaussian space, followed by synthesis, and finally transforms the synthesis result back into a new OBB. We have derived the synthesis and decoding methods, and further verified their effectiveness. The extensive experiments on several existing models show that GauS takes very little computation and improves detector’s high-precision performance. Extensive experiments verify the effectiveness, stability, and universality of the proposed algorithm. In addition, The RTMDet using GauS achieves a performance of 81.61 AP ₅₀ and gains a 0.39% improvement in mAP, which achieves the SOTA performance. Our implementation is available at: https://github.com/lzh420202/GauS.

Localization regression in oriented object detection tasks has long faced boundary discontinuity and angular discontinuity problems induced by periodic angles. These problems were successfully resolved by using a 2d Gaussian distribution to modelling the oriented bounding box (OBB). However, the angular information of square-like objects will be lost when they are converted to 2d Gaussian distribution, forming a systematic problem. Its fundamental reason is that when the aspect ratio of the object tends to 1, the equiprobability curve of 2d Gaussian distribution degenerates from an ellipse to a circle, thus losing the orientation information of the rotated object. This results in the bounding boxes of such square-like objects not being learned effectively. To resolve this problem, we used the Lamé curve (or superellipse) to modify the existing 2d Gaussian function and designed a super-Gaussian distribution. This distribution can maintain anisotropy at arbitrary aspect ratios, thus preserving the angular information of the oriented object. We used the Kullback-Leibler (KL) divergence to measure the distance between two super-Gaussian distributions and convert it into a localization loss (SGKLD) by a function. SGKLD is an improved version of KLD loss. By modifying the form of the probability distribution, we elegantly fix the angle missing problem of the traditional Gaussian distribution. We validated the effectiveness of the proposed algorithm on several datasets and obtained the performance of SOTA. Our algorithm achieves a mean average precision (mAP) of 80.07, 76.59, 62.27, and 90.55/98.13 on the DOTA-v1.0, DOTA-v1.5, DOTA-v2.0, and HRSC2016 datasets, respectively.

In general object detection tasks, accurate responses to the boundaries and classifications of potential objects are required of detectors. Given the diverse shapes of objects, these bounding boxes are typically represented as horizontal rectangles. However, in aerial imagery scenarios, horizontal bounding boxes can easily miss valid detections when objects are densely distributed. In recent years, oriented object detection algorithms have achieved considerable success. However, they focus on the boundaries of the object without taking into account its head–tail attributes and true orientation. Getting the true orientation of objects serves as the long-term aim of remote sensing object detection task. To resolve these issues, we decompose the single Gaussian distribution probability loss into a sequential dual-probability distribution, thereby establishing a one-to-one collaborative loss mechanism. This enables the model to learn the real orientation relationship from the dual distribution loss, achieving the acquisition of objects’ true orientation information. To more accurately reflect the preference of detectors, we design a new evaluation metric based on the cosine function and average precision. We conduct a series of experiments on the DOTA dataset, HRSC dataset, and FAIR1M dataset, achieving satisfactory results that validate the effectiveness of the proposed method.