No abstract

No abstract

It is difficult to detect forest fires in complex backgrounds owing to the many interfering factors in forest fire smoke. In this paper, a novel method that combines Time Domain Robust Principal Component Analysis (TRPCA) and a Two-Stream Composed of Visual Geometry Group Network (VGG) and Bi-Long Short-Term Memory (BLSTM) (TSVB) model is proposed for forest fire smoke detection. First, features are extracted from the smoke video from the spatial stream (static) and time stream (dynamic). For the spatial stream, static features are extracted from a single-frame image of the smoke video using the VGG network. For the time stream, continuous-frame binary images of the smoke are obtained using the TRPCA algorithm. Then, the dynamic features of the smoke are extracted by VGG and BLSTM. Finally, the static and dynamic features are fused using a concatenate function to achieve forest fire smoke detection. The experimental results show that compared with the single-feature model, the proposed method effectively improves learning ability and prediction ability, and shows strong robustness against interference factors in a complex background, with accuracy of forest fire smoke detection reaching 90.6%.

The target detection of smoke through remote sensing images obtained by means of unmanned aerial vehicles (UAVs) can be effective for monitoring early forest fires. However, smoke targets in UAV images are often small and difficult to detect accurately. In this paper, we use YOLOX-L as a baseline and propose a forest smoke detection network based on the parallel spatial domain attention mechanism and a small-scale transformer feature pyramid network (PDAM–STPNNet). First, to enhance the proportion of small forest fire smoke targets in the dataset, we use component stitching data enhancement to generate small forest fire smoke target images in a scaled collage. Then, to fully extract the texture features of smoke, we propose a parallel spatial domain attention mechanism (PDAM) to consider the local and global textures of smoke with symmetry. Finally, we propose a small-scale transformer feature pyramid network (STPN), which uses the transformer encoder to replace all CSP_2 blocks in turn on top of YOLOX-L’s FPN, effectively improving the model’s ability to extract small-target smoke. We validated the effectiveness of our model with recourse to a home-made dataset, the Wildfire Observers and Smoke Recognition Homepage, and the Bowfire dataset. The experiments show that our method has a better detection capability than previous methods.

No abstract

The occurrence of forest fires can lead to ecological damage, property loss, and human casualties. Current forest fire smoke detection methods do not sufficiently consider the characteristics of smoke with high transparency and no clear edges and have low detection accuracy, which cannot meet the needs of complex aerial forest fire smoke detection tasks. In this paper, we propose Dual-ResNet50-vd with SoftPool based on a recursive feature pyramid with deconvolution and dilated convolution and global optimal nonmaximum suppression (DRGNet) for high-accuracy detection of forest fire smoke. First, the Dual-ResNet50-vd module is proposed to enhance the extraction of smoke features with high transparency and no clear edges, and SoftPool is used to retain more feature information of smoke. Then, a recursive feature pyramid with deconvolution and dilated convolution (RDDFPN) is proposed to fuse shallow visual features and deep semantic information in the channel dimension to improve the accuracy of long-range aerial smoke detection. Finally, global optimal nonmaximum suppression (GO-NMS) sets the objective function to globally optimize the selection of anchor frames to adapt to the aerial photography of multiple smoke locations in forest fire scenes. The experimental results show that the DRGNet parametric number on the UAV-IoT platform is as low as 53.48 M, mAP reaches 79.03%, mAP50 reaches 90.26%, mAP75 reaches 82.35%, FPS reaches 122.5, and GFLOPs reaches 55.78. Compared with other mainstream methods, it has the advantages of real-time detection and high accuracy.

Tomato disease control is an urgent requirement in the field of intellectual agriculture, and one of the keys to it is quantitative identification and precise segmentation of tomato leaf diseases. Some diseased areas on tomato leaves are tiny and may go unnoticed during segmentation. Blurred edge also makes the segmentation accuracy poor. Based on UNet, we propose an effective image-based tomato leaf disease segmentation method called Cross-layer Attention Fusion Mechanism combined with Multi-scale Convolution Module (MC-UNet). First, a Multi-scale Convolution Module is proposed. This module obtains multiscale information about tomato disease by employing 3 convolution kernels of different sizes, and it highlights the edge feature information of tomato disease using the Squeeze-and-Excitation Module. Second, a Cross-layer Attention Fusion Mechanism is proposed. This mechanism highlights tomato leaf disease locations via gating structure and fusion operation. Then, we employ SoftPool rather than MaxPool to retain valid information on tomato leaves. Finally, we use the SeLU function appropriately to avoid network neuron dropout. We compared MC-UNet to the existing segmentation network on our self-built tomato leaf disease segmentation dataset and MC-UNet achieved 91.32% accuracy and 6.67M parameters. Our method achieves good results for tomato leaf disease segmentation, which demonstrates the effectiveness of the proposed methods.

Tomato leaf diseases have a significant impact on tomato cultivation modernization. Object detection is an important technique for disease prevention since it may collect reliable disease information. Tomato leaf diseases occur in a variety of environments, which can lead to intraclass variability and interclass similarity in the disease. Tomato plants are commonly planted in soil. When a disease occurs near the leaf’s edge, the soil backdrop in the image tends to interfere with the infected region. These problems can make tomato detection challenging. In this paper, we propose a precise image-based tomato leaf disease detection approach using PLPNet. First, a perceptual adaptive convolution module is proposed. It can effectively extract the disease’s defining characteristics. Second, a location reinforcement attention mechanism is proposed at the neck of the network. It suppresses the interference of the soil backdrop and prevents extraneous information from accessing the network’s feature fusion phase. Then, a proximity feature aggregation network with switchable atrous convolution and deconvolution is proposed by combining the mechanisms of secondary observation and feature consistency. The network solves the problem of disease interclass similarities. Finally, the experimental results show that PLPNet achieved 94.5% mean average precision with 50% thresholds (mAP50), 54.4% average recall (AR), and 25.45 frames per second (FPS) on a self-built dataset. The model is more accurate and specific for the detection of tomato leaf diseases than other popular detectors. Our proposed method may effectively improve conventional tomato leaf disease detection and provide modern tomato cultivation management with reference experience.