UAV object detection is still difficult to achieve due to large-scale variation, dense small objects, a complicated background, and resource constraints from onboard computing. To solve these problems, we develop a diffusion-enhanced detection network, DL-DEIM, tailored for aerial images. The proposed scheme generalizes the DEIM baseline across three orthogonal axes. First, we propose a lightweight backbone network called DCFNet, which utilizes a DRFD module and a FasterC3k2 module to maintain spatial information and reduce computational complexity. Second, we propose a LFDPN module, which can conduct bidirectional multi-scale fusion via frequency-spatial self-attention and deep feature refinement and largely enhance cross-scale contextual propagation for small objects. Third, we propose LAWDown, an adaptive-content-aware downsampling to preserve the discriminative representation with higher accuracy at lower resolutions, which can effectively capture the spatially-variant weights and group channel interactions. On the VisDrone2019 dataset, DL-DEIM achieves a mAP@0.5 of 34.9% and a mAP@0.5:0.95 of 20.0%, outperforming the DEIM baseline by +4.6% and +2.9%, respectively. The model maintains real-time inference speed (356 FPS) with only 4.64 M parameters and 11.73 GFLOPs. Ablation studies validate the fact that DCFNet, LFDPN, and LAWDown collaboratively contribute to the accuracy and efficiency. Visualizations also display clustered and better localized activation in crowded scenes. These results show that DL-DEIM achieves a good tradeoff between detection probability and computation burden and it can be used in practice on resource-limited UAV systems.
In this study, an improved target-detection model based on information theory is proposed to address the difficulties of crack-detection tasks, such as slender target shapes, blurred boundaries, and complex backgrounds. By introducing a multi-scale information gain mechanism and a global–local feature coupling strategy, the model has significantly improved feature extraction and expression capabilities. Experimental results show that, on a single-crack dataset, the model’s mAP@50 and mAP@50-95 are 1.6% and 0.8% higher than the baseline model RT-DETR, respectively; on a multi-crack dataset, these two indicators are improved by 1.2% and 1.0%, respectively. The proposed method shows good robustness and detection accuracy in complex scenarios, providing new ideas and technical support for in-depth research in the field of crack detection.
Multi-agent trajectory prediction is essential for autonomous driving systems, as its performance heavily depends on the quality of feature representations. This paper proposes PFR-HiVT, a lightweight and effective approach for multi-agent trajectory prediction, and evaluates it on the Argoverse 1.1 motion forecasting dataset. Although existing methods such as the Hierarchical Vector Transformer (HiVT) have achieved strong performance, they still exhibit limitations in feature extraction and feature transition across different stages of the network. To address these limitations, a collaborative feature enhancement framework is introduced, consisting of two encoder-side modules and a Progressive Feature Refinement Global Interactor (PFR-Global Interactor). Specifically, the Feature Enhancement Module (FEM) and the Attention Enhancement Module (AEM) are employed to refine local spatiotemporal features before global interaction. In addition, the PFR-Global Interactor integrates three lightweight components—the Simple Feature Refinement Module (SFR), the Lightweight Gate Module (LG), and the Residual Connection Module (RC)—to progressively refine globally interacted features prior to trajectory decoding. All proposed modules adopt lightweight designs, introducing only 230.5 k additional parameters (approximately 8.7% of the total parameters of HiVT-128). Experiments on the Argoverse 1.1 dataset show that PFR-HiVT achieves a minADE of 0.703, a minFDE of 1.041, and an MR of 0.112, outperforming the baseline HiVT model. Ablation studies further validate the effectiveness and synergy of the proposed modules.
With the rapid development of 3D sensor technology, point cloud semantic segmentation has found widespread applications in autonomous driving, remote sensing, mapping, and industrial manufacturing. However, outdoor traffic scenes present significant challenges: point clouds are inherently disordered, unevenly distributed, and unstructured. As a result, traditional point cloud semantic segmentation methods often suffer from low accuracy and unstable performance in complex tasks such as semantic segmentation and object detection. To address these limitations, this paper proposes an improved point cloud semantic segmentation method based on Cylinder3D. The proposed approach integrates the PointMamba and MS3DAM modules, which enhance the model’s ability to capture global features while preserving local details, thereby improving adaptability and recognition across multiple feature scales. Furthermore, leveraging the linear computational complexity of Mamba enables the method to maintain high efficiency when processing large-scale point cloud data. In addition, incorporating the KAT module into the encoder improves the model’s perceptual capacity and robustness in handling point clouds. Experimental results on the SemanticKITTI dataset demonstrate that the proposed method achieves a mean Intersection over Union (mIoU) of 64.98%, representing a 2.81% improvement over Cylinder3D, thereby confirming its superior segmentation accuracy compared with existing models.