Research on Target Detection and Counting Algorithms for Swarming Termites in Agricultural and Forestry Disaster Early Warning

Discriminating small moving objects within complex visual environments is a significant challenge for autonomous micro-robots that are generally limited in computational power. By exploiting their highly evolved visual systems, flying insects can effectively detect mates and track prey during rapid pursuits, even though the small targets equate to only a few pixels in their visual field. The high degree of sensitivity to small target movement is supported by a class of specialized neurons called small target motion detectors (STMDs). Existing STMD-based computational models normally comprise four sequentially arranged neural layers interconnected via feedforward loops to extract information on small target motion from raw visual inputs. However, feedback, another important regulatory circuit for motion perception, has not been investigated in the STMD pathway and its functional roles for small target motion detection are not clear. In this article, we propose an STMD-based neural network with feedback connection (feedback STMD), where the network output is temporally delayed, then fed back to the lower layers to mediate neural responses. We compare the properties of the model with and without the time-delay feedback loop and find that it shows a preference for high-velocity objects. Extensive experiments suggest that the feedback STMD achieves superior detection performance for fast-moving small targets, while significantly suppressing background false positive movements which display lower velocities. The proposed feedback model provides an effective solution in robotic visual systems for detecting fast-moving small targets that are always salient and potentially threatening.

Infrared Camera Traps (ICTs) are widely used in ecological research as a noninvasive wildlife monitoring technique, particularly for the detection and identification of animal targets. Existing ICT data screening methods face challenges in recognizing animals against complex backgrounds, particularly fast-moving or small targets. To address these issues, we proposed a target-oriented enhanced data-screening method called GFD-YOLO, which emphasized key locations in images to effectively guide the focus of the model toward target regions, thereby improving detection accuracy. We compared the effects of different preprocessing methods on detection performance. Results revealed that the proposed method improved the mean Average Precision (mAP) by 16.96%, precision by 10.13%, and recall by 24.85% compared to the YOLOv11n model. Therefore, the preprocessing method proposed in this study had significant advantages in reducing false negatives and false positives and was adaptable to wildlife detection tasks under different background conditions. In addition, this method demonstrated higher robustness in scenarios involving lighting variations and fast-moving targets.

ADD-YOLO: An algorithm for detecting animals in outdoor environments based on unmanned aerial imagery

The research of infrared image small-target detection is of great significance to security monitoring, satellite remote sensing, infrared early warning, and precision guidance systems. However, small infrared targets occupy few pixels and lack color and texture features, which make the detection of small infrared targets extremely challenging. This paper proposes an effective single-stage infrared small-target detection method based on improved FCOS (Fully Convolutional One-Stage Object Detection) and spatio-temporal features. In view of the simple features of infrared small targets and the requirement of real-time detection, based on the standard FCOS network, we propose a lightweight network model combined with traditional filtering methods, whose response for small infrared targets is enhanced, and the background response is suppressed. At the same time, in order to eliminate the influence of static noise points in the infrared image on the detection of small infrared targets, time domain features are added to the improved FCOS network in the form of image sequences, so that the network can learn the spatio-temporal correlation features in the image sequence. Finally, compared with current typical infrared small-target detection methods, the comparative experiments show that the improved FCOS method proposed in this paper had better detection accuracy and real-time performance for infrared small targets.

The technique of weak small target detection and recognition has been the key technique of the electro-optical detecting system, many scholars are engaging the research of detection for weak and small targets. The effective detection for small targets in low SNR images is still a hot research field. Because infrared sensor is easily affected by atmosphere hot radiation, long distance and sensor noise, the detected targets in infrared images often present like small targets and drowned in noise. The basic problem inherent to extent the detection range is the detection of small, low observable, no obvious structural information in images and complicated background. In order to improve the detection ratio of weak small targets and decrease the false alarming ratio in the condition of complicated background, the paper presents the technology of pretreatment of infrared images and the technology of detection for weak small targets, mainly including the technology of Sobel edge algorithm and multi-degree and multi-orientation gradient. Based upon horizon-correlative characteristic of infrared images which were gotten by scanning, considering of the target properties in complicated background, an algorithm of weak and small target detection is presented. Because the images appear horizon-correlative characteristic, Sobel horizontal operator is adopted. By this algorithm, the background clutter was suppressed. Then an adaptive threshold was proposed to extract the precise location of small target. Incorporated with the two methods, a single frame weak and small target detection algorithm was built. Its high performance was then proved in a serial of experiments. In order to solve the detection problems of weak and small infrared targets under complicated background, a detection algorithm integrating with multi-degree and multi-orientation gradient fusion is proposed. Based on the principle of infrared radiation property of target, i.e., the gradient variations of pixel gray scale in horizontal and vertical directions, the algorithm handles the property of weak and small target into analysis of image singularity and detects targets by means of multi-degree gradient step length. Then the detected results show this method can remove most false targets. As a result, the algorithms have fulfilled the engineering demands for reliability and real-time property.

Small and dim target detection is a critical challenge in hyperspectral remote sensing, particularly in complex, large-scale scenes where spectral variability across diverse land cover types complicates the detection process. In this paper, we propose a novel target reasoning algorithm named Adaptive Global Dense Nested Reasoning Network (AGDNR). This algorithm integrates spatial, spectral, and domain knowledge to enhance the detection accuracy of small and dim targets in large-scale environments and simultaneously enables reasoning about target categories. The proposed method involves three key innovations. Firstly, we develop a high-dimensional, multi-layer nested U-Net that facilitates cross-layer feature transfer, preserving high-level features of small and dim targets throughout the network. Secondly, we present a novel approach for computing physicochemical parameters, which enhances the spectral characteristics of targets while minimizing environmental interference. Thirdly, we construct a geographic knowledge graph that incorporates both target and environmental information, enabling global target reasoning and more effective detection of small targets across large-scale scenes. Experimental results on three challenging datasets show that our method outperforms state-of-the-art approaches in detection accuracy and achieves successful classification of different small targets. Consequently, the proposed method offers a robust solution for the precise detection of hyperspectral small targets in large-scale scenarios.

To address the challenges of multi-scale objects, dense distributions, occlusions, and numerous small targets in UAV image detection, we present CMS-YOLOv7, a real-time target detection method based on an enhanced YOLOv7 model. Firstly, the detection layer P2 for small targets was added to YOLOv7 to enhance the detection ability of small and medium-sized targets, and the deep detection head P5 was taken out to mitigate the influence of excessive downsampling on small target images. The anchor frame was calculated by the K-means++ method. Using the concept of Inner-IoU, the Inner-MPDIoU loss function was constructed to control the range of the auxiliary border and improve detection performance. Furthermore, the CARAFE module was introduced to replace traditional upsampling methods, offering improved integration of semantic information during the image upsampling process and enhancing feature mapping accuracy. Simultaneously, during the feature extraction stage, a non-strided convolutional SPD-Conv module was constructed using space-to-depth techniques. This module replaced certain convolutional operations to minimize the loss of fine-grained information and improve the model’s ability to extract features from small targets. Experiments on the UAV aerial photo dataset VisDrone2019 demonstrated that compared with the baseline YOLOv7 object detection algorithm, CMS-YOLOv7 achieved an improvement of 3.5% mAP@0.5, 3.0% mAP@0.5:0.95, and the number of parameters decreased by 18.54 M. The ability of small target detection was significantly enhanced.

Accurate detection of fresh jujube fruits plays a vital role in precision agriculture, enabling reliable yield estimation and supporting automation tasks such as robotic harvesting. To address the challenges of detecting such small targets (≤32 × 32 pixels) in complex orchard environments, this study proposes JFST-DETR, an efficient and robust detection model based on the Real-Time DEtection TRansformer (RT-DETR). First, to address the insufficient feature representation for small jujube fruit targets, a novel module called the Global Awareness Adaptive Module (GAAM) is designed. Building on GAAM and the innovative Spatial Coding Module (SCM), a new Spatial Enhancement Pyramid Network (SEPN) is proposed. Through the spatial-depth transformation domain and global awareness adaptive processing units, SEPN captures fine-grained features of small targets, enhancing the detection accuracy for small objects. Second, a Dynamic Sampling (DySample) operator is adopted, which optimizes feature space details via dynamic offset calculation and lightweight design, improving detection accuracy while reducing computational costs. Finally, to solve the problem of complex background interference caused by foliage occlusion and illumination variations, Pinwheel-Shaped Convolution (PSConv) is introduced. By using asymmetric padding and multi-directional convolution, PSConv enhances the robustness of feature extraction, ensuring reliable recognition in complex agricultural environments. Experimental results show that JFST-DETR achieves precision, recall, F1, mAP@50, and mAP@50:95 of 93%, 86.8%, 89.8%, 94.3%, and 75.2%. Compared to the baseline model, these metrics improve by 0.8%, 3.7%, 2.4%, 2.6%, and 3.1%, respectively. Cross-dataset evaluations further confirm its strong generalizability, demonstrating potential as a practical solution for small-target detection in intelligent horticulture.

Target detection in aerial images taken by unmanned aerial vehicles is the most widely used scene at present. Compared with ordinary images, the background of aerial images is more complex, and the target size is smaller, which results in inferior detection precision and a high false detection rate. This paper proposes a new small target detection model TCA-YOLOv5m, which is based on YOLOv5m and combines the Transformer algorithm and the Coordinate Attention (CA) mechanism. In this model, the transformer algorithm is added to the end of the backbone of the YOLOv5, which enables the model to mine more features information of images. In the neck layer of the TCA-YOLOv5m, the Path Aggregation Network (PANet) and transformer algorithm are combined to enhance the expression capacity for the feature pyramid and improve the detection precision of occluded high-density small targets, and CA is introduced to more accurately locate targets in high-density scenes. In addition, the TCA-YOLOv5m adds a detection layer to improve the ability to capture small targets. This paper uses VisDrone 2019 as experimental data, and takes experiments to compare the detection precision and detection speed of the proposed model with baseline models. The experiment results indicate that the detection precision of the TCA-YOLOv5m reaches 97.4%, which is 5.2% higher than that of YOLOv5; the value of MAP @ 50 reaches 58.5%, which is 14.8% higher than YOLOv5. The Frames Per Second (FPS) of the TCA-YOLOv5m is 12.96 f/s, which ensures a certain real-time performance. Therefore, the TCA-YOLOv5m is suitable for the task of detecting dense small targets in aerial images.

The infrared small target’s detection and tracking are important parts of the automatic target recognition. When the camera platform equipped with an infrared camera moves, the small target’s position change in the imaging plane is affected by the composite motion of the small target and the camera platform. Traditional detection and tracking algorithms may lose the small target and make the follow-up detection and tracking fail because of not considering the camera platform’s movement. Moreover, when there exist small targets with different motion features in the camera’s view, some detection and tracking algorithms can’t recognize different targets based on their motion features because there are no trajectories in a unified coordinate system, which may lead to the true small targets undetected or detected incorrectly . To solve those problems, we present a method under the condition of moving camera platform. Firstly, get the camera platform’s motion information from the inertial measurement values, and then decouple to remove the motion of the camera platform itself by means of coordinate transformation. Next, estimate the trajectories of the small targets with different motion features based on their position changes in the same imaging plane coordinate system. Finally, recognize different small targets preliminarily based on their different trajectories. Experimental results show that this method can improve the small target’s detection probability. Furthermore, when the camera platform fails to track the small target, it’s possible to predict the position of the small target in the next frame based on the fitted motion equation and realize sustained and stable tracking.

The identification of small land targets in remote sensing imagery has emerged as a significant research objective. Despite significant advancements in object detection strategies based on deep learning for visible remote sensing images, the performance of detecting a small and densely distributed number of small targets remains suboptimal. To address this issue, this study introduces an improved model named YOLOV4_CPSBi, based on the YOLOV4 architecture, specifically designed to enhance the detection capability of small land targets in remote sensing imagery. The proposed model enhances the traditional CSPNet by redefining its channel partitioning and integrating this enhanced structure into the neck part of the YOLO network model. Additionally, the conventional pyramid fusion structure used in the traditional BiFPN is removed. By integrating a weight-based bidirectional multi-scale mechanism for feature fusion, the model is capable of effectively reasoning about objects of various sizes, with a particular focus on detecting small land targets, without introducing a significant increase in computational costs. Using the DOTA dataset as research data, this study quantifies the object detection performance of the proposed model. Compared with various baseline models, for the detection of small targets, its AP performance has been improved by nearly 8% compared with YOLOV4. By combining these modifications, the proposed model demonstrates promising results in identifying small land targets in visible remote sensing images.

Infrared small target (IRST) detection focuses on segmenting small infrared targets from complex backgrounds. Recent Convolutional Neural Networks (CNNs) show strong performance on detecting infrared small targets with complex background. Existing CNNs-based methods mainly have two weaknesses. First, features of small targets are likely to lose in deep stages of networks. Second, infrared small targets are always shapeless, which will cause more false detections. To solve the above mentioned two weaknesses, we propose a saliency-transformer combined knowledge distillation guided network (ST-KDNet). In our proposed ST-KDNet, we first use transformer-based segmentation branch to extract the attention region of small targets. Then we apply saliency detection branch to filter some irrelevant similar targets, where the saliency mask is used to guide the transformer-based segmentation branch. To further enhance representation ability of small target on the low-level feature, we introduce a knowledge distillation guidance. Extensive experiments on benchmark datasets, MDFA and SIRST, prove that ST-KDNet outperforms previous state-of-the-art (SOTA) methods.KeywordsST-KDNetSaliency-transformerKnowledge distillationInfrared small target detection

Target detection in unmanned aerial vehicle application scenarios has other problems, such as dense targets. The existing unmanned aerial vehicle target detection model with high computational complexity makes it difficult to meet real-time unmanned aerial vehicle target detection, and the detection accuracy of small targets is low. To address these problems, we propose an improved YOLOv7 small target detection model based on context and pyramidal attention that can cope with dense unmanned aerial vehicle scenarios - CPA-YOLOv7. This model embeds our proposed lightweight multi-scale attentional feature spatial pyramid pooling module, which can better distinguish between small and large target features, reducing the computational effort while improving the detection accuracy of the model. Secondly, we design a contextual dynamic fusion attention module in the network to fuse global and local contextual information and dynamically assign features to multiple groups of channels; in the multi-scale fusion process, it effectively increases the characterization ability of small target features and enables the network to better focus on small target information. Finally, we improve Wise-Intersection-over-Union loss as the regression loss function, add a moderation factor to retain some of the high and low-quality sample weights to improve the regression accuracy of high-quality anchor frames, and use the dynamic non-monotonic focusing mechanism to increase the model's focus on ordinary quality anchor frames to improve the model's localization performance and robustness to low-quality samples. Numerous experimental results show that on the unmanned aerial vehicle datasets VisDrone2021-DET and AI-TOD, the mAP values of our model are 2.3% and 1.1% higher than those of the YOLOv7 model with fewer parameters introduced, and the computational speed reaches 146 frames per second (FPS), which can meet the real-time requirements of unmanned aerial vehicle detection.

Small maritime targets, e.g., periscope tubes, jet skies, swimmers and small boats, are potential threats for naval ships under many conditions, but are difficult to detect with current radar systems due to their limited radar cross section and the presence of sea clutter. On the other hand, applications of lidar systems have shown that the reflections from small targets are significantly stronger than reflections from the sea surface. As a result, dedicated lidar systems are potential tools for the detection of small maritime targets. A geometric approach is used to compare the diffuse reflection properties of cylinders and spheres with flat surfaces, which is used to estimate the maximum detectable range of such objects for a given lidar system. Experimental results using lasers operating at 1.06 μm and 1.57 μm confirm this theory and are discussed. Small buoys near Scheveningen harbor could be detected under adverse weather over more than 9 km. Extrapolation of these results indicates that small targets can be detected out to ranges of approximately 20 km.

Conventional scan-to-scan integration correlation (SIC) algorithms can detect small and stationary targets. However, they are ineffective in detecting small and fast-moving targets. This paper presents an improved SIC algorithm together with clutter suppression measures that remove or decrease sea clutter. The algorithm divides the scan-to-scan integration (SI) into two branches, one provides optimum clutter attenuation by means of SI weighting while the other ensures that targets are detected even if they are fast-moving. Sea clutter suppression can lower detection thre-sholds and, at the same time, increase signal-to-clutter ratio. Simulation results show that the proposed approach greatly improves the detection capability for warship radar.