Lightweight Pest Object Detection Model for Complex Economic Forest Tree Scenarios
Open Access
Abstract
Pest control in economic forests is a crucial aspect of sustainable forest resource management, yet it faces bottlenecks such as low efficiency and high miss rates for small objects. Based on the RT-DETR model, this paper proposes LightFAD-DETR, a lightweight architecture integrated with feature aggregation diffusion, designed for complex economic forest scenarios. Firstly, by employing the YOLOv9 lightweight backbone network to compress the computational base load, we introduce the RepNCSPELAN4-CAA module, which integrates re-parameterization techniques and one-dimensional strip convolution. This enhances the model’s ability for cross-regional modeling of slender insect morphologies. Secondly, a feature aggregation diffusion network is designed, incorporating a dimension-aware selective integration mechanism. This dynamically fuses shallow detail features with deep semantic features, effectively mitigating information loss for small objects occluded by foliage. Finally, a re-parameterized batch normalization technique is introduced to reconstruct the AIFI module. Combined with a progressive training strategy, this eliminates redundant parameters, thereby enhancing inference efficiency on edge devices. Experimental validation demonstrates that compared to the baseline RT-DETR model, LightFAD-DETR achieves a 1.4% improvement in , while reducing parameters by 41.7% and computational load by 35.0%. With an inference speed reaching 106.3 FPS, the method achieves balanced improvements in both accuracy and lightweight design.
Keywords
Funding Information
- Key R & D projects of the Guangxi Science and Technology Program (Guike.AB24010338, GuikeZY22096012, 32360374, GXRDCF202307-01)
- Central Government Guides Local Science and Technology Development Fund Projects (Guike.AB24010338, GuikeZY22096012, 32360374, GXRDCF202307-01)
- National Natural Science Foundation of China (Guike.AB24010338, GuikeZY22096012, 32360374, GXRDCF202307-01)
- Independent Research Project (Guike.AB24010338, GuikeZY22096012, 32360374, GXRDCF202307-01)
This publication has 49 references indexed in Scilit:
- Pest-PVT: A model for multi-class and dense pest detection and counting in field-scale environmentsComputers and Electronics in Agriculture, 2024
- YOLOv9: Learning What You Want to Learn Using Programmable Gradient Information2024
- Detecting emerald ash borer boring vibrations using an encoder‐decoder and improved DenseNet modelPest Management Science, 2024
- Machine learning for pest detection and infestation prediction: A comprehensive reviewWIREs Data Mining and Knowledge Discovery, 2024
- AC-YOLO: Multi-category and high-precision detection model for stored grain pests based on integrated multiple attention mechanismsExpert Systems with Applications, 2024
- DsP-YOLO: An anchor-free network with DsPAN for small object detection of multiscale defectsExpert Systems with Applications, 2023
- Present status of insecticide impacts and eco-friendly approaches for remediation-a reviewEnvironmental Research, 2023
- MD-YOLO: Multi-scale Dense YOLO for small target pest detectionComputers and Electronics in Agriculture, 2023
- Diseases and insect pests challenge to meet wood production demand of Tectona grandis (L.), a high-value tropical tree speciesJournal of Plant Diseases and Protection, 2023
- Tree insect pests and pathogens: a global systematic review of their impacts in urban areasUrban Ecosystems, 2023