Zhenyu Liu

Efficient monitoring of pork freshness is essential to minimize spoilage-related losses in the meat industry. To address the limitations of existing detection technologies, namely high cost, poor timeliness and high environmental sensitivity, this study developed a novel electronic nose system integrating a hybrid sensor array with dynamic gas path control. By combining metal oxide semiconductor (MOS) and electrochemical sensors (e.g., MQ137, MQ136), the system exhibits high sensitivity to the key volatile organic compounds (VOCs) released during pork spoilage, achieving a detection accuracy of over 90% in identifying spoilage stages. Combined with a dual-mode gas circuit design (solenoid valve switching time: 0.85 s), the reliability of the system was further demonstrated. This technology offers an economical and efficient real-time monitoring solution for slaughterhouses and cold chain logistics, providing a new low-cost scientific approach for pork freshness assessment.

Intensive pig farming presents substantial challenges for individual health monitoring due to high stocking densities, complex occlusion scenarios, and the need for continuous real-time surveillance. Existing monitoring approaches rely heavily on manual inspection, which is labor-intensive and prone to delayed detection of abnormal behaviors and disease symptoms. This study proposes an embedded intelligent monitoring system integrating a pan-tilt gimbal platform with an improved multi-object tracking and anomaly detection framework for automated pig health surveillance. The system employs a modified Periodfill_DeepSORT algorithm that incorporates a ReID network with appearance features and motion prediction trajectories to maintain identity consistency under occlusion and re-entry scenarios. For anomaly detection, a lightweight YOLOv8-based network was trained on 772 abnormal samples across three behavioral categories: movement abnormalities, postural abnormalities, and disease-related abnormalities. Experimental results demonstrate that the Periodfill_DeepSORT algorithm achieves a Multiple Object Tracking Accuracy (MOTA) of 95.34%, a Multiple Object Tracking Precision (MOTP) of 94.77%, and an IDF1 score of 96.88%, with only 12 identity switches across 2000 frames involving 12 targets—27 fewer than the standard DeepSORT algorithm. In occlusion scenarios, MOTA improved from 61.1% to 78.3%. The anomaly detection network achieves an overall detection accuracy of 94.5%, representing an 8.8 percentage point improvement over the baseline model, with recognition accuracies of 96.2% for movement abnormalities, 94.1% for postural abnormalities, and 92.8% for disease-related abnormalities. The system operates at 90 frames per second on embedded hardware with a power consumption of 3.2 watts and a startup time of approximately 1 s, with gimbal angle errors maintained within 3°. These results demonstrate the system’s effectiveness and practical feasibility for real-time intelligent health monitoring in intensive livestock farming environments.

Improving animal welfare through standardized management protocols remains a key challenge in intensive pig production. Auditory enrichment, such as music, represents a promising non-invasive strategy, yet its application is often empirical, lacking mechanistic understanding and objective assessment tools. This study investigated growing pigs’ active preferences for structured musical parameters to establish a precision auditory enrichment framework. Seventy-two crossbred pigs were subjected to a free-choice paradigm under simulated farm conditions, with a 2 × 2 factorial design manipulating musical stimulus type (a guqin string piece vs. a Mozart wind excerpt) and tempo (fast: 200 bpm vs. slow: 65 bpm) was continuously quantified using an enhanced YOLOv11-based automated recognition system (mean average precision mAP₅₀: 90.5% ± 1.5%). Results revealed highly parameter-dependent effects: the slow-tempo GS stimulus and the fast-tempo MF stimulus significantly prolonged occupancy time (p < 0.01) and elicited distinct profiles. The GS stimulus promoted a calm, investigative state, increasing lying, exploration, and drinking time (p < 0.05), while the MF stimulus stimulated an active playful state, characterized by increased walking and playing (p < 0.05). Other musical combinations showed negligible effects, whereas noise exposure consistently triggered stress-related responses. This study establishes an integrated “parametric design → automated assessment → specific output” methodology for precision auditory enrichment, providing an empirical basis for evidence-based acoustic protocols in commercial pig production.

To address the low efficiency of traditional pig behavior monitoring methods, this study proposes a swine motion information recognition algorithm and develops a corresponding monitoring system. The system adopts a front-end/back-end separated architecture. The front-end provides video playback and control, multi-target identification, and trajectory visualization. The back-end performs motion detection and background modeling using the MOG2 algorithm and generates trajectory heatmaps through DBSCAN-based clustering. Two operational workflows are supported, namely manual annotation and automatic feature extraction. The system calculates key motion parameters, including velocity and momentum, and enables the export of the processed data. Experimental results demonstrate that the proposed system can effectively analyze swine motion characteristics and trajectory information, providing an accurate and efficient monitoring solution for large-scale pig farming, with practical value for optimizing husbandry management and improving animal health.

Accidental crushing by sows is the primary cause of pre-weaning piglet mortality in intensive production, often due to the spatiotemporal lag of manual inspection. While Internet of Things (IoT) solutions exist, they frequently face challenges such as vision occlusion, high hardware costs, and latency. To address these, this study developed a low-cost multi-modal edge computing system based on TinyML. Using an ESP32-S3 microcontroller, the system employs a “Motion-Gated Acoustic Detection” strategy, activating a lightweight 1D-CNN model to identify piglet screams only when an IMU detects high-risk postural transitions of the sow. Results show the quantized model (5.1 KB) achieves 95.56% accuracy and 2 ms inference latency. The total end-to-end response latency is within 179 ms, ensuring intervention within the early “golden rescue window.” The low-power design enables the battery life to cover the entire lactation period. Field tests demonstrated that the system intercepted identified crushing risks within the monitored cohort, supporting its potential for significantly improving piglet survival probability. This research overcomes the limitations of single-modal monitoring and provides a scalable, cost-effective engineering intervention for enhancing animal welfare and achieving intelligent, unattended supervision in precision livestock farming.

This study explores the intervention effect of Sparassis Latifolia Polysaccharides (SLPs) on abnormal glucose metabolism in C57BL/6J mice, induced by a high-fat diet and streptozotocin (STZ). The results show that high-concentration SLPs significantly improve abnormal glucose metabolism. A mouse facial expression dataset was constructed, covering five glucose metabolic states: Norm, Pre-stage of Abnormal Glucose Metabolism, Abnormal Glucose Metabolism, Early Stage of SLPs Intervention, and Late Stage of SLPs Intervention, based on fasting blood glucose and facial expression features. To achieve non-invasive detection of abnormal glucose metabolism, a lightweight deep learning model, LFPP-YOLO, was proposed. This model incorporates a Partial Self-Attention (PSA) module to enhance global context information extraction and utilizes the L-FFCA structure for multi-level feature enhancement and background suppression, improving its adaptability to complex expressions. The PIoU v2 loss function was employed to optimize facial expression localization accuracy and robustness. Experimental results show that the LFPP-YOLO model achieves an average facial detection accuracy of 95.1% across the five metabolic states, with a compact size of 2.4 MB and a real-time inference speed of 5ms, significantly outperforming existing models. This model provides a novel approach and technical support for non-invasive screening and personalized intervention of abnormal glucose metabolism in mice.