In this study, an improved African vulture optimization algorithm (IAVOA) that combines the African vulture optimization algorithm (AVOA) with both quasi-oppositional learning and differential evolution is proposed to address specific drawbacks of the AVOA, including low population diversity, bad development capability, and unbalanced exploration and development capabilities. The improved algorithm has three parts. First, quasi-oppositional learning is introduced in the population initialization and exploration stages to improve population diversity. Second, a differential evolution operator is introduced in the local search position update of each population to improve exploration capability. Third, adaptive parameters are introduced to the differential evolution operator, thus balancing the algorithm exploration and development. A numerical simulation experiment based on 36 different types of benchmark functions showed that while the IAVOA can enhance the convergence speed and solution accuracy of the basic AVOA and two variants of AVOA, IAVOA outperforms the other 7 swarm intelligence algorithms in the mean and best values of 33 benchmark functions.

Though Synthetic Aperture Radar (SAR) Automatic Target Recognition (ATR) via Convolutional Neural Networks (CNNs) has made huge progress toward deep learning, some key issues still remain unsolved due to the lack of sufficient samples and robust model. In this paper, we proposed an efficient transferred Max-Slice CNN (MS-CNN) with L²-Regularization for SAR ATR, which could enrich the features and recognize the targets with superior performance. Firstly, the data amplification method is presented to reduce the computational time and enrich the raw features of SAR targets. Secondly, the proposed MS-CNN framework with L²-Regularization is trained to extract robust features, in which the L²-Regularization is incorporated to avoid the overfitting phenomenon and further optimizing our proposed model. Thirdly, transfer learning is introduced to enhance the feature representation and discrimination, which could boost the performance and robustness of the proposed model on small samples. Finally, various activation functions and dropout strategies are evaluated for further improving recognition performance. Extensive experiments demonstrated that our proposed method could not only outperform other state-of-the-art methods on the public and extended MSTAR dataset but also obtain good performance on the random small datasets.

Remote sensing change detection (RSCD) aims to identify changes within bi-temporal registered images. However, existing deep learning-based RSCD networks often suffer from large numbers of parameters, high computational complexity, and low inference speed, making it challenging to achieve efficient inference in real-world deployments. Additionally, current models lack robust feature fitting capabilities, necessitating the development of an efficient and powerful RSCD model to address this issue. Therefore, we propose a novel RSCD network named Efficient Adjacent Feature Harmonizer Network (EAFH-Net) with fast computational speed and lightweight design. It is based on MobileNetV2, considering that change maps of different sizes contain temporal information of bitemporal features and spatial information at various scales, we introduce Multiscale Feature Neighbor Fusion Module (MFNFM) to address the lack of interaction between sophisticated-level and elementary-level features, and Spatial and Channel Harmonizer Module (SCFHM) to harmonize the spatiotemporal information of the change maps. Moreover, data-driven deep learning (DL) algorithms face another challenge due to insufficient granularity and the need for more practical datasets. Therefore, we present UAV-CD+, a dataset comprising 2002 pairs of bi-temporal UAV low-altitude images, each sized at 1024×1024. We performed experiments on three publicly accessible datasets in conjunction with UAV-CD+, comparing the results with other state-of-the-art (SOTA) methods. EAFH-Net attains the utmost precision, obtaining 91.74% on LEVIR-CD, 84.28% on SYSU-CD, 95.07% on WHU-CD, 79.12% on CLCD, and 70.12% on UAV-CD+. We have our model code available at the following link: https://github.com/yikuizhai/UCSFH-Net.

Steel Surface Defect (SSD) recognition plays a critical role in guaranteeing the quality and economic benefits of industrial metal products. At present, deep learning-based recognition methods have achieved excellent performance within related fileds. However, these methods usually depend on the availability of extensive training data and label information. For SSD recognition, the SSD data is often insufficient and labels are scarce, making it hard to train a supervised deep model. In this article, we proposed a novel few-shot SSD recognition method based on self-supervised constrastive learning (Self-SCL), which can effectively learn data representation with unlabeled data in the pretraining stage and learn categories from a few labeled samples in the fine-tuning stage. In the method, we proposed a information-guided Teacher-Student (TS) pretrained model with Min-Max Instances Similarity (MMIS) regularization. In the pretraining stage, we utilized the multi-crop mechanism to build a Teacher Encoder (TE) with multi-view information, and leveraged the TE to guide the Student Encoder (SE) with information. Meanwhile, the SE also provided the TE with “questions” during the update process, which the SE also shared updated information with the TE for further refinement. Specially, we introduced MMIS to address the problem of ambiguity between intra-class and inter-class distances in SSD data. During the fine-tuning phase, we fine-tuned the pretrained model obtained by TSMMIS with a few labeled data, and tested the SSD recognition accuracy in four SSD datasets. As the experimental results indicate, our method significantly outperforms other state-of-the-art methods on the four SSD datasets.

As samples of steel defects are industrially limited, it is challenging for most deep learning methods that rely on ample labeled data to identify steel surface defects. Recently, contrastive learning has achieved good performance in natural image classification tasks with few labeled samples, yet two obstacles prevent its effective application to steel surface defect images. One is that due to the presence of inter-class and intra-class similar samples in steel surface defect, the fixed contrast strength in contrastive learning will destroy the potential semantic information of defect samples. Another is that contrastive learning requires a large amount of unlabeled data, whereas steel surface defect samples are insufficient. To overcome the above-mentioned problems, a novel framework named flexible and diverse contrastive learning (FDCL) is proposed. This framework consists of two parts, flexible contrast (FiCo) and diverse generative adversarial networks (DGANs). Diverse images generated by the DGAN and real images are fed into FiCo for representation learning. In the FiCo, the contrast strength among samples is flexibly adjusted by the proposed variable temperature discrimination and feature reconstruction (FR). In addition, the output features (OF) of FiCo will be used as input to the DGAN generator to improve image quality, thus further facilitating representation learning. The proposed FDCL is implemented on four standard steel surface defect data sets, and the experimental results demonstrated that it achieves superior performance over state-of-the-art methods. Our code is available at: https://github.com-/jiacongc/FDCL .

The application of UAVs is becoming increasingly extensive. However, high-precision autonomous landing is still a major industry difficulty. The current algorithm is not well-adapted to light changes, scale transformations, complex backgrounds, etc. To address the above difficulties, a deep learning method was here introduced into target detection and an attention mechanism was incorporated into YOLOX; thus, a UAV positioning algorithm called attention-based YOLOX (A-YOLOX) is proposed. Firstly, a novel visual positioning pattern was designed to facilitate the algorithm’s use for detection and localization; then, a UAV visual positioning database (UAV-VPD) was built through actual data collection and data augmentation and the A-YOLOX model detector developed; finally, corresponding high- and low-altitude visual positioning algorithms were designed for high- and low-altitude positioning logics. The experimental results in the actual environment showed that the AP50 of the proposed algorithm could reach 95.5%, the detection speed was 53.7 frames per second, and the actual landing error was within 5 cm, which meets the practical application requirements for automatic UAV landing.