To suppress the spread of COVID-19, accurate diagnosis at an early stage is crucial, chest screening with radiography imaging plays an important role in addition to the real-time reverse transcriptase polymerase chain reaction (RT-PCR) swab test. Due to the limited data, existing models suffer from incapable feature extraction and poor network convergence and optimization. Accordingly, a multi-stage residual network, MSRCovXNet, is proposed for effective detection of COVID-19 from chest x-ray (CXR) images. As a shallow yet effective classifier with the ResNet-18 as the feature extractor, MSRCovXNet is optimized by fusing two proposed feature enhancement modules (FEM), i.e., low-level and high-level feature maps (LLFMs and HLFMs), which contain respectively more local information and rich semantic information, respectively. For effective fusion of these two features, a single-stage FEM (MSFEM) and a multi-stage FEM (MSFEM) are proposed to enhance the semantic feature representation of the LLFMs and the local feature representation of the HLFMs, respectively. Without ensembling other deep learning models, our MSRCovXNet has a precision of 98.9% and a recall of 94% in detection of COVID-19, which outperforms several state-of-the-art models. When evaluated on the COVIDGR dataset, an average accuracy of 82.2% is achieved, leading other methods by at least 1.2%.

Oriented object detection in remote sensing is a critical task for accurately location and measurement of the interested targets. Despite of its success in object detection, deep learning-based detectors rely heavily on extensive data annotation. However, variations in object appearance significantly increase the difficulty and the cost of creating large-scale annotated datasets. Semi-supervised learning aims to utilize unlabeled data to enhance object detectors. Among these, pseudo-label-based methods have shown promising results recently. Nonetheless, as training progresses, the accumulation of errors in pseudo-labels leads to prediction bias without corrections. To tackle this particular challenge, we present a semi-supervised learning pipeline, named “Dual Teacher”, for improving the reliability of pseudo labels in the semi-supervised oriented object detection. Firstly, to mitigate the bias caused by limited annotated data, a global burn-in strategy is introduced at the beginning of training, which guides the student detector to learn the feature extraction on a global scale. Additionally, an online bounding box correction module is proposed to decrease the occurrence of mislabeled instances and enhance the reliability of detection. These improvements are facilitated by an additional detector, instead of a single teacher model in the teacher-student architecture. Dual Teacher reduces the dependency on the quality of pseudo-labels related to the model complexity, and combines the strengths of both the two-stage and one-stage detectors. With only 20% labeled data, Dual Teacher outperforms fully supervised R-FCOS, YOLOX-s and R-RCNN by up to 2% on both DOTA and SODA-A datasets. This reveals its potential in reducing labor-intensive tasks and enhancing robustness against environmental interference and noisy labels. The code is available at https://github.com/ZYFFF-CV/DualTeacher-semisup.git.

An oriented bounding box (OBB) is preferable over a horizontal bounding box (HBB) in accurate object detection. Most of existing works utilize a two-stage detector for locating the HBB and OBB, respectively, which have suffered from the misaligned horizontal proposals and the interference from complex backgrounds. To tackle these issues, region of interest transformer and attention models were proposed, yet they are extremely computationally intensive. To this end, we propose a semi-anchor-free detector (SAFDet) for object detection in aerial images, where a rotation-anchor-free-branch (RAFB) is used to enhance the foreground features via precisely regressing the OBB. Meanwhile, a center-prediction-module (CPM) is introduced for enhancing object localization and suppressing the background noise. Both RAFB and CPM are deployed during training, avoiding increased computational cost of inference. By evaluating on DOTA and HRSC2016 datasets, the efficacy of our approach has been fully validated for a good balance between the accuracy and computational cost.