Insect Identification in the Wild: The AMI Dataset

Lecture notes in computer science, Год журнала: 2024, Номер unknown, С. 55 - 73

Опубликована: Дек. 1, 2024

Язык: Английский

---

A novel dataset and deep learning object detection benchmark for grapevine pest surveillance

Frontiers in Plant Science, Год журнала: 2024, Номер 15

Опубликована: Дек. 12, 2024

Flavescence dorée (FD) poses a significant threat to grapevine health, with the American leafhopper, Scaphoideus titanus , serving as primary vector. FD is responsible for yield losses and high production costs due mandatory insecticide treatments, infected plant uprooting, replanting. Another potential vector mosaic Orientus ishidae commonly found in agroecosystems. The current monitoring approach, which involves periodic human identification of yellow sticky traps, labor-intensive time-consuming. Therefore, there compelling need develop an automatic pest detection system leveraging recent advances computer vision deep learning techniques. However, progress developing such has been hindered by lack effective datasets training. To fill this gap, our study contributes fully annotated dataset S. O. from includes more than 600 images, approximately 1500 identifications per class. Assisted entomologists, we performed annotation process, trained, compared performance two state-of-the-art object algorithms: YOLOv8 Faster R-CNN. Pre-processing, including cropping eliminate irrelevant background information image enhancements improve overall quality dataset, was employed. Additionally, tested impact altering resolution data augmentation, while also addressing issues related class detection. results, evaluated through 10-fold cross validation, revealed promising accuracy, achieving [email protected] 92%, F1-score above 90%, mAP@[0.5:0.95] 66%. Meanwhile, R-CNN reached 86% 55%, respectively. This outcome offers encouraging prospects management strategies fight against dorée.

Язык: Английский

---

In-field monitoring of ground-nesting insect aggregations using a scaleable multi-camera system

Ecological Informatics, Год журнала: 2025, Номер unknown, С. 103004 - 103004

Опубликована: Янв. 1, 2025

Язык: Английский

---

Advanced Insect Detection Network for UAV-Based Biodiversity Monitoring

Remote Sensing, Год журнала: 2025, Номер 17(6), С. 962 - 962

Опубликована: Март 9, 2025

The Advanced Insect Detection Network (AIDN), which represents a significant advancement in the application of deep learning for ecological monitoring, is specifically designed to enhance accuracy and efficiency insect detection from unmanned aerial vehicle (UAV) imagery. Utilizing novel architecture that incorporates advanced activation normalization techniques, multi-scale feature fusion, custom-tailored loss function, AIDN addresses unique challenges posed by small size, high mobility, diverse backgrounds insects images. In comprehensive testing against established models, demonstrated superior performance, achieving 92% precision, 88% recall, an F1-score 90%, mean Average Precision (mAP) score 89%. These results signify substantial improvement over traditional models such as YOLO v4, SSD, Faster R-CNN, typically show performance metrics approximately 10–15% lower across similar tests. practical implications AIDNs are profound, offering benefits agricultural management biodiversity conservation. By automating classification processes, reduces labor-intensive tasks manual enabling more frequent accurate data collection. This collection quality frequency enhances decision making pest conservation, leading effective interventions strategies. AIDN’s design capabilities set new standard field, promising scalable solutions UAV-based monitoring. Its ongoing development expected integrate additional sensory real-time adaptive further applicability, ensuring its role transformative tool monitoring environmental science.

Язык: Английский

---

Attention mechanism‐based ultralightweight deep learning method for automated multi‐fruit disease recognition system

Agronomy Journal, Год журнала: 2025, Номер 117(2)

Опубликована: Март 1, 2025

Abstract Automated disease recognition plays a pivotal role in advancing smart artificial intelligence (AI)‐based agriculture and is crucial for achieving higher crop yields. Although substantial research has been conducted on deep learning‐based automated plant systems, these efforts have predominantly focused leaf diseases while neglecting affecting fruits. We propose an efficient architecture effective fruit with state‐of‐the‐art performance to address this gap. Our method integrates advanced techniques, such as multi‐head attention mechanisms lightweight convolutions, enhance both efficiency performance. Its ultralightweight design emphasizes minimizing computational costs, ensuring compatibility memory‐constrained edge devices, enhancing accessibility practical usability. Experimental evaluations were three diverse datasets containing multi‐class images of disease‐affected healthy samples sugar apple ( Annona squamosa ), pomegranate Punica granatum guava Psidium guajava ). proposed model attained exceptional results test set accuracies weighted precision, recall, f1‐scores exceeding 99%, which also outperformed pretrain large‐scale models. Combining high accuracy represents significant step forward developing accessible AI solutions agriculture, contributing the advancement sustainable agriculture.

Язык: Английский

---