YOLO-WL: A Lightweight and Efficient Framework for UAV-Based Wildlife Detection.

Area of Science:

• Computer Vision
• Ecological Monitoring
• Artificial Intelligence

Background:

• Accurate wildlife detection in Unmanned Aerial Vehicle (UAV) imagery is vital for biodiversity conservation.
• Challenges include species similarity, environmental interference, and small target sizes.

Purpose of the Study:

• Introduce YOLO-WL, a specialized wildlife detection algorithm for UAV-based monitoring.
• Enhance detection accuracy and robustness for diverse ecological environments.

Main Methods:

• Developed a Multi-Scale Dilated Depthwise Separable Convolution (MSDDSC) module for enriched semantic representation.
• Integrated a Multi-Scale Large Kernel Spatial Attention (MLKSA) mechanism to focus on animal regions.
• Employed a Shallow-Spatial Alignment Path Aggregation Network (SSA-PAN) with Spatial Guidance Fusion (SGF) for precise feature fusion.

Main Results:

• YOLO-WL achieved 94.2% mAP@0.5 and 58.0% mAP@0.5:0.95 on the WAID dataset, outperforming state-of-the-art methods.
• Demonstrated robustness and generalization on Aerial Sheep and AI-TOD datasets.
• Successfully addressed challenges of species similarity, environmental disturbances, and small target detection.

Conclusions:

• YOLO-WL is an effective tool for improving UAV-based wildlife monitoring.
• The algorithm supports enhanced ecological conservation practices through accurate wildlife detection.