Birds-YOLO: A Bird Detection Model for Dongting Lake Based on Modified YOLOv11

TL;DR

This paper introduces Birds-YOLO, an improved bird detection model for Dongting Lake, which performs better than the original YOLOv11 in complex natural environments.

Contribution

The paper proposes Birds-YOLO, a modified YOLOv11 model with enhanced feature extraction and multi-scale detection for bird monitoring in complex environments.

Findings

Birds-YOLO achieves a 5.0% improvement in recall and 3.5% increase in mAP@0.5 on the CUB200-2011 dataset.

On the DTH-Birds dataset, it gains 3.7% in precision, 3.7% in recall, and 2.6% in mAP@0.5.

The model shows strong generalization and robustness in both public and private benchmarks.

Abstract

Detecting birds in natural environments like Dongting Lake is challenging due to cluttered backgrounds, birds of different sizes, and a wide variety of species. This study aimed to create a more accurate and efficient model to detect birds in such complex settings. We collected real-world images of 47 bird species from Dongting Lake and developed an improved detection system that better recognizes birds in various conditions. Our model improves the way it focuses on important details and enhances the ability to identify birds of different sizes. Tests show that our model performs significantly better than the basic version, with higher success rates in detecting birds correctly while missing fewer birds. It works well on both public bird image collections and our own field data. This advancement helps create more reliable tools for monitoring bird populations in the wild, which is important for protecting biodiversity and understanding ecosystem health. The improved model can support conservation efforts by enabling automated, real-time bird observation in natural habitats, making ecological research more efficient and scalable.

To address the challenges posed by complex background interference, varying target sizes, and high species diversity in bird detection tasks in the Dongting Lake region, this paper proposes an enhanced bird detection model named Birds-YOLO, based on the YOLOv11 framework. First, the EMA mechanism is introduced to replace the original C2PSA module. This mechanism synchronously captures global dependencies in the channel dimension and local detailed features in the spatial dimension, thereby enhancing the model’s robustness in cluttered environments. Second, the model incorporates an improved RepNCSPELAN4-ECO module, by reasonably integrating depthwise separable convolution modules and combining them with an adaptive channel compression mechanism, to strengthen feature extraction and multi-scale feature fusion, effectively enhances the detection capability for bird targets at different scales. Finally, the neck component of the network is redesigned using lightweight GSConv convolution, which integrates the principles of grouped and spatial convolutions. This design preserves the feature modeling capacity of standard convolution while incorporating the computational efficiency of depthwise separable convolution, thereby reducing model complexity without sacrificing accuracy. Experimental results show that, compared to the baseline YOLOv11n, Birds-YOLO achieves a 5.0% improvement in recall and a 3.5% increase in mAP@0.5 on the CUB200-2011 dataset. On the in-house DTH-Birds dataset, it gains 3.7% in precision, 3.7% in recall, and 2.6% in mAP@0.5, demonstrating consistent performance enhancement across both public and private benchmarks. The model’s generalization ability and robustness are further validated through extensive ablation studies and comparative experiments, indicating its strong potential for practical deployment in bird detection tasks in complex natural environments such as Dongting Lake.

Figures