Forestpest-YOLO: A High-Performance Detection Framework for Small Forestry Pests

2025/10/01 by Aoduo Li, Li, Aoduo, P. M. Lin +11
Agricultural and Biological Sciences · Biochemistry, Genetics and Molecular Biology · Environmental Science · #Artificial Intelligence (cs.AI) #Computer Vision and Pattern Recognition (cs.CV) #Date Palm Research Studies #FOS: Computer and information sciences #Forest Insect Ecology and Management #Insect and Arachnid Ecology and Behavior

paper · pdf · doi:10.48550/arxiv.2510.00547

openalex publication_date 2025/10/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Detecting agricultural pests in complex forestry environments using remote sensing imagery is fundamental for ecological preservation, yet it is severely hampered by practical challenges. Targets are often minuscule, heavily occluded, and visually similar to the cluttered background, causing conventional object detection models to falter due to the loss of fine-grained features and an inability to handle extreme data imbalance. To overcome these obstacles, this paper introduces Forestpest-YOLO, a detection framework meticulously optimized for the nuances of forestry remote sensing. Building upon the YOLOv8 architecture, our framework introduces a synergistic trio of innovations. We first integrate a lossless downsampling module, SPD-Conv, to ensure that critical high-resolution details of small targets are preserved throughout the network. This is complemented by a novel cross-stage feature fusion block, CSPOK, which dynamically enhances multi-scale feature representation while suppressing background noise. Finally, we employ VarifocalLoss to refine the training objective, compelling the model to focus on high-quality and hard-to-classify samples. Extensive experiments on our challenging, self-constructed ForestPest dataset demonstrate that Forestpest-YOLO achieves state-of-the-art performance, showing marked improvements in detecting small, occluded pests and significantly outperforming established baseline models.

Citations

• MCA-LLaVA: Manhattan Causal Attention for Reducing Hallucination in Large Vision-Language Models
• MAC-Lookup: Multi-Axis Conditional Lookup Model for Underwater Image Enhancement
• LLM-Enhanced Multimodal Fusion for Cross-Domain Sequential Recommendation
• High-Fidelity Functional Ultrasound Reconstruction via A Visual Auto-Regressive Framework
• Underwater Image Restoration Through a Prior Guided Hybrid Sense Approach and Extensive Benchmark Analysis
• Image Fusion for Cross-Domain Sequential Recommendation
• Image Augmentation Agent for Weakly Supervised Semantic Segmentation
• Underwater Image Restoration via Polymorphic Large Kernel CNNs
• Prompt Categories Cluster for Weakly Supervised Semantic Segmentation
• Seeing Clearly by Layer Two: Enhancing Attention Heads to Alleviate Hallucination in LVLMs
• High-Fidelity Document Stain Removal via A Large-Scale Real-World Dataset and A Memory-Augmented Transformer
• Test-Time Intensity Consistency Adaptation for Shadow Detection
• ForgeryTTT: Zero-Shot Image Manipulation Localization with Test-Time Training
• Instance-adaptive Zero-shot Chain-of-Thought Prompting
• GRE²-MDCL: Graph Representation Embedding Enhanced via Multidimensional Contrastive Learning
• Lagrange Duality and Compound Multi-Attention Transformer for Semi-Supervised Medical Image Segmentation
• SMAFormer: Synergistic Multi-Attention Transformer for Medical Image Segmentation
• Dual-Hybrid Attention Network for Specular Highlight Removal
• Depth-aware Test-Time Training for Zero-shot Video Object Segmentation
• MedPrompt: Cross-Modal Prompting for Multi-Task Medical Image Translation
• Efficient Remote Sensing Segmentation With Generative Adversarial Transformer
• High-Resolution Document Shadow Removal via A Large-Scale Real-World Dataset and A Frequency-Aware Shadow Erasing Net
• Devignet: High-Resolution Vignetting Removal via a Dual Aggregated Fusion Transformer With Adaptive Channel Expansion
• DocDeshadower: Frequency-Aware Transformer for Document Shadow Removal
• Explicit Visual Prompting for Low-Level Structure Segmentations
• CoordFill: Efficient High-Resolution Image Inpainting via Parameterized Coordinate Querying
• Brain Diffuser: An End-to-End Brain Image to Brain Network Pipeline
• A Large-scale Film Style Dataset for Learning Multi-frequency Driven Film Enhancement
• No More Strided Convolutions or Pooling: A New CNN Building Block for Low-Resolution Images and Small Objects
• YOLOX: Exceeding YOLO Series in 2021
• Simple Copy-Paste is a Strong Data Augmentation Method for Instance Segmentation
• An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
• VarifocalNet: An IoU-aware Dense Object Detector
• Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detection
• End-to-End Object Detection with Transformers
• Small Object Detection using Context and Attention
• Dynamic Convolution: Attention over Convolution Kernels
• Bridging the Gap Between Anchor-based and Anchor-free Detection via Adaptive Training Sample Selection
• Bridging the Gap Between Anchor-Based and Anchor-Free Detection via Adaptive Training Sample Selection
• EfficientDet: Scalable and Efficient Object Detection
• Distance-IoU Loss: Faster and Better Learning for Bounding Box Regression
• Distance-IoU Loss: Faster and Better Learning for Bounding Box Regression
• Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression
• CBAM: Convolutional Block Attention Module
• Path Aggregation Network for Instance Segmentation
• Non-local Neural Networks
• Squeeze-and-Excitation Networks
• Perceptual Generative Adversarial Networks for Small Object Detection
• Deformable Convolutional Networks
• Feature Pyramid Networks for Object Detection
• From Redundancy to Relevance: Information Flow in LVLMs Across Reasoning Tasks

Related

• FID-Net: A Feature-Enhanced Deep Learning Network for Forest Infestation Detection
• PDT: Uav Target Detection Dataset for Pests and Diseases Tree
• MSFNet-CPD: Multi-Scale Cross-Modal Fusion Network for Crop Pest Detection
• YOLO-RS: Remote Sensing Enhanced Crop Detection Methods
• A Multimodal Approach for Advanced Pest Detection and Classification