Unmanned Aerial Vehicles (UAVs) equipped with onboard processors enable edge intelligence for smart-grid Internet of Things (IoT) inspection, yet insulator defect detection from aerial imagery remains challenging due to cluttered backgrounds, small targets, and strict latency constraints. Compared with handcrafted methods that rely on manual feature design and parameter tuning, this paper presents an end-to-end improved Faster R-CNN detector whose feature extractor is redesigned via convolutional feature reuse and a two-layer multi-scale fusion scheme with a deeper extraction path, preserving fine-grained cues while limiting model complexity. Experiments on the Chinese Power Line Insulator Dataset (CPLID) and a newly constructed Glass Insulator Dataset (GID) achieve [email protected] values of 97.75% and 96.79%, respectively, which is 7.1% higher than a standard Faster R-CNN (VGG16) baseline on CPLID while reducing the model size to 58 MB. On a UAV system (DJI M300 + H20T + Jetson Xavier NX), TensorRT acceleration runs at 8.5 FPS, outperforming Faster R-CNN (3.2 FPS) and YOLOv11 (5.3 FPS) under the same edge setting. Outdoor flight tests on 330 kV transmission lines further validate field readiness, supporting low-latency and bandwidth-efficient UAV inspection for smart-grid IoT maintenance.

In this paper, a novel early forest fire detection and localization system is proposed based on an unmanned aerial vehicle (UAV) equipped with RGB and thermal cameras. The forest fire detection is achieved by the proposed bimodal detection network. Fire localization, i.e., fire distance estimation, is realized by the proposed monocular depth estimation network and scale recovery strategy. To validate the effectiveness of the proposed method, the performance of the detection network is tested using the bimodal dataset UAVFire presented in this paper and the public dataset FLAME2. The proposed depth estimation network is validated on the public dataset WildUAV. The experimental results show that the proposed bimodal detection network has a AP of 97.9% and 98.2% on the UAVFire and FLAME2 datasets, respectively, with a model parameter of 6.5 MB. Compared with other unimodal detection methods as well as bimodal detection methods, the detection method in this paper performs well in terms of both the detection accuracy and the model lightweighting. The model parameter of the proposed monocular depth estimation network is only 0.69 MB, and the comparison results with other methods on theWildUAV dataset prove that the depth estimation method in this paper achieves a good balance between accuracy and model lightweighting. Finally, the detection and distance estimation of early small fires are validated on the dataset UAVFire, demonstrating the effectiveness of the proposed method.

This paper presents a lightweight method for detecting forest fires using dual-modal remote sensing images captured by an unmanned aerial vehicle (UAV). The aim is to achieve efficient fire monitoring on a computationally resource-constrained UAV platform. The proposed detection network is based on the improved YOLOv8, which uses RGB image and thermal image as network input at the same time. A lightweight dual-modal feature fusion module named DFM is designed to effectively combine RGB and thermal features. The existing C2f module in YOLOv8 was replaced by the lightweight module C2f-F, along with the addition of the parameter-free attention module SimAM. This improvement improves the detection performance of the model while minimizing the model parameters. The evaluation experimental results on the FLAME 2 dataset show that the accuracy of the proposed dual-modal forest fire detection method reaches 98.4%, and the model size is only 2.9MB, which achieves a good balance between accuracy and number of parameters compared with other mainstream methods. Additionally, on the iCrest 2-s edge computing device, the detection speed reaches 20.67FPS, further confirming that this lightweight approach satisfies the real-time detection requirements for forest fires.

The samples representing abnormal situation is usually very few in the dataset, which makes it difficult to learn the features of abnormal samples by machine-learning-based methods. To improve the accuracy of anomaly detection, the number of abnormal samples should be expanded to ensure the balance of the dataset. In this paper, a discrete synthetic minority oversampling technique (D-SMOTE) is proposed to generate new samples. A closed area is constructed using the three nearest abnormal samples in the dataset. The new samples are then uniformly interpolated in a closed area. By this means, the problem of the imbalance for the original dataset is handled, thus improving the data quality. Based on the expanded datasets, a two-dimensional convolutional neural network (2D CNN) is constructed to detect abnormal samples. In experiments, three cases and different machine learning methods are considered for comparison. Several indexes including accuracy, precision, confusion matrix, F1-score, and Recall have been used to evaluate the detection effectiveness. The results show that the abnormal samples can be detected accurately using oversampling data obtained from the proposed D-SMOTE method.