Related Experiment Video
Updated: Jun 18, 2025
06:19
Low-Cost Automated Flight Intercept Trap for the Temporal Sub-Sampling of Flying Insects Attracted to Artificial Light at Night
Published on: December 29, 2021
2.5K
Using Data Augmentation to Improve the Generalization Capability of an Object Detector on Remote-Sensed Insect Trap
Jozsef Suto1,2
1Department of Informatics Systems and Networks, Faculty of Informatics, University of Debrecen, Kassai Street, 26, 4028 Debrecen, Hungary.
Sensors (Basel, Switzerland)
|July 27, 2024
Summary
Automating insect pest detection requires accurate models, but data is scarce. This study enhanced insect trap image datasets using data augmentation, significantly improving detection accuracy for the YOLOv5 model.
Area of Science:
- Agricultural Entomology
- Computer Vision
- Machine Learning
Background:
- Traditional insect monitoring relies on manual inspection of sticky traps, which is labor-intensive and inefficient.
- Automated insect pest detection using computer vision models is crucial for modern agriculture, but faces challenges due to limited training data.
Purpose of the Study:
- To investigate the impact of various data augmentation techniques on the performance of insect pest detection models.
- To evaluate the effectiveness of combining geometric, photometric, and mosaic augmentation strategies for remote-sensed insect trap images.
Main Methods:
- Systematic application of geometric and photometric data augmentation techniques, individually and in combination, to two distinct insect trap image datasets.
- Utilized the YOLOv5 (small) object detection model to assess performance improvements.
- Integrated photometric transformations with mosaic augmentation for enhanced results.
Main Results:
- The combination of augmentation methods significantly increased the model's mean average precision (mAP50) from 0.844 to 0.992 and 0.421 to 0.727 across two datasets.
- Integrating photometric transformations into mosaic augmentation further boosted mAP50 values to 0.999 and 0.756, outperforming native augmentation combinations.
Conclusions:
- Data augmentation is a critical strategy to overcome data scarcity in insect pest detection.
- Optimized combinations of augmentation techniques, particularly those integrating photometric transformations with mosaic augmentation, substantially enhance the accuracy of YOLOv5 for automated insect monitoring.
