Ships comprise the only and most important ocean transportation mode. Thus, ship detection is one of the most critical technologies in ship monitoring, which plays an essential role in maintaining marine safety. Optical remote-sensing images contain rich color and texture information, which is beneficial to ship detection. However, few optical remote-sensing datasets are open publicly due to the issue of sensitive data and copyrights, and only the HRSC2016 dataset is built for the ship-detection task. Moreover, almost all general object detectors suffer from the failure of multi-scale ship detection because of the diversity of spatial resolution and ship size. In this paper, we re-annotate the HRSC2016 dataset and supplement 610 optical remote-sensing images to build a new open source ship-detection benchmark dataset with rich multi-scale ship objects named the HRSC2016-MS dataset. In addition, we further explore the potential of a recursive mechanism in the field of object detection and propose a novel multi-scale ship-detection framework (MSSDet) in optical remote-sensing images. The success of detecting multi-scale objects depends on the hierarchical pyramid structure in the object-detection framework. However, the inherent semantic and spatial gaps among hierarchical pyramid levels seriously affect detection performance. To alleviate this problem, we propose a joint recursive feature pyramid (JRFP), which can generate semantically strong and spatially refined multi-scale features. Extensive experiments were conducted on the HRSC2016-MS, HRSC2016, and DIOR datasets. Detailed ablation studies directly demonstrated the effectiveness of the proposed JRFP architecture and also showed that the proposed method has excellent generalizability. Comparisons with state-of-the-art methods showed that the proposed method achieves competitive performance, i.e., 77.3%, 95.8%, and 73.3% mean average precision accuracy on the HRSC2016-MS, HRSC2016, and DIOR datasets, respectively.

Unmanned aerial vehicles (UAVs) have attracted increasing attention in recent years because of their broad range of applications in city security, military reconnaissance, disaster rescue, and so on. As one of the critical algorithms in the field of artificial intelligence, object tracking greatly improves the working efficiency of UAVs. However, unmanned aerial vehicle (UAV) object tracking still faces many challenges. UAV objects provide limited textures and contours for feature extraction due to their small sizes. Moreover, to capture objects continuously, a UAV camera must constantly move with the object. The above two reasons are usual causes of object-tracking failures. To this end, we propose an end-to-end discriminative tracker called TMDiMP. Inspired by the self-attention mechanism in Transformer, a novel memory-aware attention mechanism is embedded into TMDiMP, which can generate discriminative features of small objects and overcome the object-forgetting problem after camera motion. We also build a UAV object-tracking dataset with various object categories and attributes, named VIPUOTB, which consists of many video sequences collected in urban scenes. Our VIPUOTB is different from other existing datasets in terms of object size, camera motion speed, location distribution, etc. TMDiMP achieves competitive results on our VIPUOTB dataset and three public datasets, UAVDT, UAV123, and VisDrone, compared with state-of-the-art methods, thus demonstrating the effectiveness and robustness of our proposed algorithm.

Deep image compression and text-to-image generation represent two distinct paradigms in visual representation learning: one focuses on coded representations, while the other emphasizes semantic representations. This paper bridges this gap within a unified framework that integrates image generation and compression. We propose Joint Semantic and Coded Generation for Conditional Latent Coding (JSCCL), which combines minimal coded features with semantic descriptions from vision-language models to guide a controllable rectified flow model. This joint guidance enables high-fidelity reference generation capturing both semantic coherence and structural details. Through iterative alignment in the latent domain, we synthesize conditional priors for efficient transformer-based encoding and decoding. Experimental results on Kodak, CLIC, and Tecnick datasets demonstrate up to 15.8% BD-rate improvement over VVC with only 0.01 bpp overhead. Unlike existing generative compression methods approximating inputs solely at the semantic level, our approach achieves both semantic understanding and pixel-level reconstruction.