Unmanned aerial vehicle (UAV) aerial image object detection is a valuable and challenging research field. Despite the breakthrough of deep learning-based object detection networks in natural scenes, UAV images often exhibit characteristics such as a high proportion of small objects, dense distribution, and significant variations in object scales, posing great challenges for accurate detection. To address these issues, we propose an innovative multi-scale feature fusion small object detection network (MFFSODNet). First, concerning the high proportion of small objects in UAV images, an additional tiny object prediction head is introduced instead of the large object prediction head. This approach provides a good detection accuracy of small objects and significantly reduces the parameters. Secondly, to enhance the feature extraction capability of the network for fine-grained information from small objects, a multi-scale feature extraction module (MSFEM) is designed, which could extract rich and valuable multi-scale feature information through convolution operation of different scales on multiple branches. Thirdly, to fuse the fine-grained information from shallow feature maps and the semantic information from deep feature maps, a new bidirectional dense feature pyramid network (BDFPN) is proposed. By expanding the feature pyramid network scale and introducing skip connections, BDFPN achieves efficient multi-scale information fusion. Extensive experiments on the VisDrone and UAVDT benchmark datasets demonstrate that MFFSODNet outperforms state-of-the-art object detection methods and further validate the effectiveness and generalization of MFFSODNet on photovoltaic array defect datasets (PVD).

Photovoltaic panels are the core components of photovoltaic power generation systems, and their quality directly affects power generation efficiency and circuit safety. To address the shortcomings of existing photovoltaic defect detection technologies, such as high labor costs, large workloads, high sensor failure rates, low reliability, high false alarm rates, high network demands, and slow detection speeds of traditional algorithms, we propose an algorithm named ST-YOLO specifically for photovoltaic module defect detection. This algorithm is based on YOLOv8s. First, it introduces the C2f-SCconv convolution module, which is based on SCconv convolution. This module reduces the computational burden of model parameters and improves detection speed through lightweight design. Additionally, the Triplet Attention mechanism is incorporated, significantly enhancing detection accuracy without substantially increasing model parameter computations. Experiments on a self-built photovoltaic array infrared defect image dataset show that ST-YOLO, compared to the baseline YOLOv8s, achieves a 15% reduction in model weight, a 2.9% improvement in Precision, and a 1.4% increase in mAP@0.5. Compared to YOLOv7-Tiny and YOLOv5s, ST-YOLO also demonstrates superior detection performance and advantages. This indicates that ST-YOLO has significant application value in photovoltaic defect detection.

Unmanned aerial vehicles (UAVs) have been widely used in aerial photography and target detection tasks due to their flexibility and unique perspective. However, small targets often suffer from insufficient resolution, uneven scale distribution, and complex background clutter, which are constrained by imaging conditions such as high-altitude imaging, long-distance capture, and wide field of view. These factors weaken the feature representation and generalization ability of the model, becoming the key bottleneck that restricts the improvement of small target detection accuracy in UAV scenarios. To address the above issues, this paper proposes a small target detection algorithm for UAV perspective, namely MTD-YOLO. First, a Parallel Multi-Scale Receptive Field Unit (PMSRFU) is designed. This unit effectively enhances the receptive field range of feature extraction and the fusion ability of multi-scale contextual information by introducing parallel branches with different-sized convolutional kernels. Second, we embed PMSRFU into a C2f block to form C2f-PMSRFU, which reuses shallow details and fuses multi-scale features to clarify edges and textures in small targets, yielding stronger fine-grained representations. Finally, an efficient detection head with task decoupling, dynamic alignment, and adaptive scale adjustment capabilities, namely SDIDA-Head, is proposed, which significantly improves the model’s small target detection accuracy. Extensive experiments on the VisDrone2019 and HazyDet datasets demonstrate that MTD-YOLO achieves a 7.6% and 6.6% increase in mAP@0.5 compared to the baseline YOLOv8n, respectively. Meanwhile, the Precision is improved by 6.0% and 1.1%, and the Recall is enhanced by 7.5% and 6.9%, respectively. These results fully validate the effectiveness and superiority of the proposed method in UAV small target detection tasks.

Salient object detection (SOD) is a key preprocessing step in computer vision, widely used in object tracking, action recognition, and image retrieval, among other fields. However, traditional SOD algorithms often face issues such as rough object boundaries, incomplete extraction of global image features, and insufficient attention to key areas. To address these problems, an improved significance object detection network — CGDINet (Coordinate Attention-Group Consensus Aggregation Module-Depth Auxiliary Module-Inverse Saliency Pyramid Reconstruction Network)—is proposed. CGDINet introduces the Group Consensus Aggregation Module (GCAM) embedded with the Coordinate Attention (CA) mechanism, named CAGM (Coordinate Attention-Group Consensus Aggregation Module), to enhance global feature capture capabilities and improve the processing of directional features. Additionally, the Depth Auxiliary Module (DAM) is incorporated to enhance the focus on important regions. Experiments were conducted on five public datasets (DUT-OMRON, ECSSD, PASCAL-S, DUTS-TE, and HKU-IS). The results show that CGDINet outperforms other mainstream significance object detection models in evaluation metrics such as maxF _β , S _α , and MAE, with almost no increase in computational cost (FLOPs) and parameters. The experimental results validate that CGDINet can effectively address the issues of incomplete global feature extraction and insufficient attention to key areas, thereby significantly enhancing the performance of significance object detection.