A refined YOLOv8A-SD model is introduced to address pig detection challenges in aerial surveillance of pig farms. The model incorporates the ADown attention mechanism and a dual-task strategy combining detection and segmentation tasks. Testing was conducted using top-view footage from a large-scale pig farm in Sichuan, with 924 images for detection training, 216 for validation, and 2985 images for segmentation training, with 1512 for validation. The model achieved 96.1% Precision and 96.3% mAP50 in detection tasks while maintaining strong segmentation performance (IoU: 83.1%). A key finding reveals that training with original images while applying segmentation preprocessing during testing provides optimal results, achieving exceptional counting accuracy (25.05 vs. actual 25.09 pigs) and simplifying practical deployment. The research demonstrates YOLOv8A-SD’s effectiveness in complex farming environments, providing reliable monitoring capabilities for intelligent farm management applications.

Maize (Zea mays L.), a vital global food crop, relies on its stomatal structure for regulating photosynthesis and responding to drought. Conventional manual stomatal detection methods are inefficient, subjective, and inadequate for high-throughput plant phenotyping research. To address this, we curated a dataset of over 1500 maize leaf epidermal stomata images and developed a novel lightweight detection model, StomaYOLO, tailored for small stomatal targets and subtle features in microscopic images. Leveraging the YOLOv11 framework, StomaYOLO integrates the Small Object Detection layer P2, the dynamic convolution module, and exploits large-scale epidermal cell features to enhance stomatal recognition through auxiliary training. Our model achieved a remarkable 91.8% mean average precision (mAP) and 98.5% precision, surpassing numerous mainstream detection models while maintaining computational efficiency. Ablation and comparative analyses demonstrated that the Small Object Detection layer, dynamic convolutional module, multi-task training, and knowledge distillation strategies substantially enhanced detection performance. Integrating all four strategies yielded a nearly 9% mAP improvement over the baseline model, with computational complexity under 8.4 GFLOPS. Our findings underscore the superior detection capabilities of StomaYOLO compared to existing methods, offering a cost-effective solution that is suitable for practical implementation. This study presents a valuable tool for maize stomatal phenotyping, supporting crop breeding and smart agriculture advancements.