Dairy cow behavior carries important health information. Timely and accurate detection of behaviors such as drinking, feeding, lying, and standing is meaningful for monitoring individual cows and herd management. In this study, a model called Res-DenseYOLO is proposed for accurately detecting the individual behavior of dairy cows living in cowsheds. Specifically, a dense module was integrated into the backbone network of YOLOv5 to strengthen feature extraction for actual cowshed environments. A CoordAtt attention mechanism and SioU loss function were added to enhance feature learning and training convergence. Multi-scale detection heads were designed to improve small target detection. The model was trained and tested on 5516 images collected from monitoring videos of a dairy cowshed. The experimental results showed that the performance of Res-DenseYOLO proposed in this paper is better than that of Fast-RCNN, SSD, YOLOv4, YOLOv7, and other detection models in terms of precision, recall, and mAP metrics. Specifically, Res-DenseYOLO achieved 94.7% precision, 91.2% recall, and 96.3% mAP, outperforming the baseline YOLOv5 model by 0.7%, 4.2%, and 3.7%, respectively. This research developed a useful solution for real-time and accurate detection of dairy cow behaviors with video monitoring only, providing valuable behavioral data for animal welfare and production management.

Effective and comprehensive evaluation of cold stress is critical for healthy dairy cow breeding in the winter. Previous studies on dairy cow cold stress have considered thermal environmental factors but not physiological factors or air quality. Therefore, this study aimed to propose a multilevel fuzzy comprehensive evaluation (FCE) method for cold stress in dairy cows based on the analytic hierarchy process (AHP) and a genetic algorithm (GA). First, the AHP was used to construct an evaluation index system for cold stress in dairy cows from 3 dimensions: thermal environment (temperature, relative humidity, wind speed, and illumination), physiological factors (respiratory rate, body surface temperature), and air quality [NH₃, CO₂, inhalable particulate matter (PM₁₀)]. Second, the consistency test of the judgment matrix was transformed into a nonlinear constrained optimization problem and solved using the GA. Next, based on fuzzy set theory, the comment set and membership function were established to classify the degree of cold stress into 5 levels: none, mild, moderate, high, and extreme. Then, the degree of cold stress in cows was obtained using multilevel fuzzy comprehensive judgment. To investigate the effect of illumination indicators on cold stress in dairy cows, 24 prelactation cows from the south and north sides were selected for a 117-d comprehensive cold stress evaluation. The results showed that the mean mild cold stress durations were 605.3 h (25.22 d) and 725.5 h (30.23 d) and the moderate cold stress durations were 67.2 h (2.8 d) and 96 h (4.0 d) on the south and north sides, respectively. Simultaneously, generalized linear mixed model showed that there were significant correlations between the daily cold stress duration and milk yield, feeding time, lying time, and active steps in the cows on both sides. This method can reasonably indicate cow cold stress conditions and better guide cold protection practices in actual production.

Volatile fatty acids (VFAs) and methane are the main products of rumen fermentation. Quantitative studies of rumen fermentation parameters can be performed using in vitro techniques and machine learning methods. The currently proposed models suffer from poor generalization ability due to the small number of samples. In this study, a prediction model for rumen fermentation parameters (methane, acetic acid (AA), and propionic acid (PA)) of dairy cows is established using the stacking ensemble learning method and in vitro techniques. Four factors related to the nutrient level of total mixed rations (TMRs) are selected as inputs to the model: neutral detergent fiber (NDF), acid detergent fiber (ADF), crude protein (CP), and dry matter (DM). The comparison of the prediction results of the stacking model and base learners shows that the stacking ensemble learning method has better prediction results for rumen methane (coefficient of determination ($R^2$) = 0.928, root mean square error (RMSE) = 0.968 mL/g), AA ($R^2$ = 0.888, RMSE = 1.975 mmol/L) and PA ($R^2$ = 0.924, RMSE = 0.74 mmol/L). And the stacking model simulates the variation of methane and VFAs in relation to the dietary fiber content. To demonstrate the robustness of the model in the case of small samples, an independent validation experiment was conducted. The stacking model successfully simulated the transition of rumen fermentation type and the change of methane content under different concentrate-to-forage (C:F) ratios of TMR. These results suggest that the rumen fermentation parameter prediction model can be used as a decision-making basis for the optimization of dairy cow diet compositions, rapid screening of methane emission reduction, feed beneficial to dairy cow health, and improvement of feed utilization.

The products quality and welfare of animals are closely related to the environment parameters in livestock buildings. A monitoring and control method of environment parameters in livestock buildings based on wireless sensor network is proposed in this paper. Temperature, humidity, light, carbon dioxide concentration, ammonia concentration, and hydrogen sulfide concentration can be monitored in real time by this method. The above six parameters will be adjusted and controlled through WLS algorithm and the minimizing deviation criteria. Compared with the traditional method, the high labor cost will be saved and energy consumption will be decreased. The experimental results show that this method can in real time monitor, effectively adjust, and control the environmental parameters in livestock buildings.

In real-world dairy farming environments, object recognition models often suffer from missed or false detections due to complex backgrounds and cow occlusions. In response to these issues, this paper proposes FSCA-YOLO, a multi-object cow behavior recognition model based on an improved YOLOv11 framework. First, the FEM-SCAM module is introduced along with the CoordAtt mechanism to enable the model to better focus on effective behavioral features of cows while suppressing irrelevant background information. Second, a small object detection head is added to enhance the model’s ability to recognize cow behaviors occurring at the distant regions of the camera’s field of view. Finally, the original loss function is replaced with the SIoU loss function to improve recognition accuracy and accelerate model convergence. Experimental results show that compared with mainstream object detection models, the improved YOLOv11 in this section demonstrates superior performance in terms of precision, recall, and mean average precision (mAP), achieving 95.7% precision, 92.1% recall, and 94.5% mAP—an improvement of 1.6%, 1.8%, and 2.1%, respectively, over the baseline YOLOv11 model. FSCA-YOLO can accurately extract cow features in real farming environments, providing a reliable vision-based solution for cow behavior recognition. To support specific behavior recognition and in-region counting needs in multi-object cow behavior recognition and tracking systems, OpenCV is integrated with the recognition model, enabling users to meet the diverse behavior identification requirements in groups of cows and improving the model’s adaptability and practical utility.

Cow-YOLO: Automatic cow mounting detection based on non-local CSPDarknet53 and multiscale neck