DCNN for Pig Vocalization and Non-Vocalization Classification: Evaluate Model Robustness with New Data.

Since pig vocalization is an important indicator of monitoring pig conditions, pig vocalization detection and recognition using deep learning play a crucial role in the management and welfare of modern pig livestock farming. However, collecting pig sound data for deep learning model training takes time and effort. Acknowledging the challenges of collecting pig sound data for model training, this study introduces a deep convolutional neural network (DCNN) architecture for pig vocalization and non-vocalization classification with a real pig farm dataset. Various audio feature extraction methods were evaluated individually to compare the performance differences, including Mel-frequency cepstral coefficients (MFCC), Mel-spectrogram, Chroma, and Tonnetz. This study proposes a novel feature extraction method called Mixed-MMCT to improve the classification accuracy by integrating MFCC, Mel-spectrogram, Chroma, and Tonnetz features. These feature extraction methods were applied to extract relevant features from the pig sound dataset for input into a deep learning network. For the experiment, three datasets were collected from three actual pig farms: Nias, Gimje, and Jeongeup. Each dataset consists of 4000 WAV files (2000 pig vocalization and 2000 pig non-vocalization) with a duration of three seconds. Various audio data augmentation techniques are utilized in the training set to improve the model performance and generalization, including pitch-shifting, time-shifting, time-stretching, and background-noising. In this study, the performance of the predictive deep learning model was assessed using the k-fold cross-validation (k = 5) technique on each dataset. By conducting rigorous experiments, Mixed-MMCT showed superior accuracy on Nias, Gimje, and Jeongeup, with rates of 99.50%, 99.56%, and 99.67%, respectively. Robustness experiments were performed to prove the effectiveness of the model by using two farm datasets as a training set and a farm as a testing set. The average performance of the Mixed-MMCT in terms of accuracy, precision, recall, and F1-score reached rates of 95.67%, 96.25%, 95.68%, and 95.96%, respectively. All results demonstrate that the proposed Mixed-MMCT feature extraction method outperforms other methods regarding pig vocalization and non-vocalization classification in real pig livestock farming.

Ammonia and Particulate Matter Emissions at a Korean Commercial Pig Farm and Influencing Factors.

Quantifying emission factors of ammonia and particulate matter (PM) in livestock production systems is crucial for assessing and mitigating the environmental impact of animal production and for ensuring industry sustainability. This study aimed to determine emission factors of ammonia, total suspended particles (TSPs), PM10, and PM2.5 for piglets and growing-finishing pigs at a commercial pig farm in Korea. It also sought to identify factors influencing these emission factors. The research found that the emission factors measured were generally lower than those currently used in Korea, but were consistent with findings from individual research studies in the literature. Seasonal variations were observed, with ammonia emissions peaking in spring and autumn, and PM emissions rising in summer. Correlation analyses indicated that the number of animals and their average age correlated positively with both ammonia and PM emission factors. Ventilation rate was also positively correlated with PM emissions. Future extended field measurements across diverse pig farms will offer deeper insights into the emission factors of pig farms in Korea, guiding the development of sustainable livestock management practices.

Evaluation of Actual Ventilation Rates and Efficiency in Research-Scale Pig Houses Based on Ventilation Configurations.

Accurate ventilation control is crucial for maintaining a healthy and productive environment in research-specialized pig facilities. This study aimed to evaluate actual ventilation rates and ventilation efficiency by investigating different inlet and exhaust configurations. The research was conducted in two pig rooms, namely pig room A and pig room B, in the absence of animals and workers to focus solely on evaluating the ventilation system's performance. Actual ventilation rates were measured using hood-type anemometers, and the local air change per hour was analyzed at various measurement points via tracer gas decay experiments. The results demonstrated that specific inlet and exhaust combinations, such as side inlet/chimney outlet and ceiling inlet/side outlet, exhibited higher ventilation rates. However, the measured ventilation rates were much lower than the manufacturer's specifications. The side exhaust fan closer to the pig activity space demonstrated better ventilation effectiveness for the animals than the chimney exhaust fan. Additionally, the ceiling inlet exhibited superior air distribution and uniformity. Lower ventilation rates and higher infiltration ratios resulted in reduced ventilation efficiency, with the difference between pig and worker activity spaces being pronounced. This study emphasizes the importance of selecting optimal inlet and exhaust configurations to achieve efficient ventilation and create a healthy environment for both pigs and workers.

Behavioral changes of sows with changes in flattening rate.

In this study, considering the difficulties for all farms to convert farm styles to animal welfare-based housing, an experiment was performed to observe the changes in the behavior and welfare of sows when the slat floor was changed to a collective breeding ground. Twenty-eight sows used in this study were between the second and fifth parities to minimize the influence of parity. Using a flats floor cover, the flattening rates were treated as 0%, 20%, 30%, 40%, and 50%. Data collection was the behavior of sows visually observed using a camera (e.g., standing, lying, fighting and excessive biting behaviors, and abnormal behaviors) and the animal welfare level measured through field visits. Lying behavior was found to be higher (p < 0.01) as the flattening rate increased, and sows lying on the slatted cover also increased as the flattening rate increased (p < 0.01). Fighting behavior wasincreased when the flattening rate was increased to 20%, and chewing behavior was increased (p < 0.05) as the flattening rate increased. The animal welfare level of sows, 'good feeding', it was found that all treatment groups for body condition score and water were good at 100 (p < 0.05). 'Good housing' was the maximum value (100) in each treatment group. As the percentage of floor increased, the minimum good housing was increased from 78 in 0% flattening rate to 96 in 50% flattening rate. The maximum (100) 'good health' was achieved in the 0% and 20% flattening rates, and it was 98, 98, and 99 in the 30%, 50%, and 40% flattening rate, respectively. 'Appropriate behavior' score was significantly lower than that of other paremeters, but when the flattening ratio was 0% and 20%, the maximum and minimum values were 10. At 40% and 50%, the maximum values were 39 and 49, respectively, and the minimum values were analyzed as 19 for both 40% and 50%. These results will be used as basic data about sow welfare for farmers to successfully transition to group housing and flat floors.