Related Experiment Video
Updated: May 28, 2026
08:16
Collecting and Processing Drone-based Remotely Sensed Data for Use in Forest Recovery Monitoring
Published on: October 24, 2025
RCAF-Net: Wildlife Target Detection in Complex Forest Scenarios.
Xiuling Yu1, Chenxiao Qu1, Yifu Xu1
1College of Information and Technology, Jilin Agricultural University, Changchun 130118, China.
Animals : an Open Access Journal From MDPI
|May 27, 2026
Summary
This study introduces RCAF-Net, an improved wildlife target detection model for complex forests. It enhances detection accuracy and efficiency for automated wildlife monitoring.
Area of Science:
- Computer Vision
- Wildlife Ecology
- Machine Learning
Background:
- Wildlife monitoring in complex forest environments faces challenges like background interference, occlusion, and scale variation.
- Detecting distant animals as small targets with limited detail is difficult.
- Efficient deployment is crucial for practical wildlife monitoring systems.
Purpose of the Study:
- To develop an improved wildlife target detection model, RCAF-Net, based on YOLO11n.
- To enhance shallow feature representation, multi-scale contextual modeling, and feature fusion.
- To introduce a lightweight detection head for balancing accuracy and computational cost.
Main Methods:
- Proposed RCAF-Net model based on YOLO11n architecture.
- Enhancements include improved feature representation, contextual modeling, and feature fusion.
- Incorporated a lightweight detection head for efficiency.
Main Results:
- RCAF-Net achieved Precision (89.3%), Recall (78.4%), mAP@0.5 (87.3%), and mAP@0.5:0.95 (67.3%) on a custom dataset.
- Demonstrated improved generalization performance on the Wildlife Computer Vision Model dataset.
- Achieved approximately 27 FPS on the Jetson TX2 NX platform.
Conclusions:
- RCAF-Net significantly improves wildlife target detection in challenging forest environments.
- The model offers a balance between high detection accuracy and computational efficiency.
- RCAF-Net shows potential for automated wildlife monitoring applications.
