The complex multi-stage process of meat processing encompasses critical phases, including slaughtering, cooling, cutting, packaging, warehousing, and logistics. The quality and nutritional value of the final meat product are significantly influenced by each processing link. To address the major challenges in the meat processing industry, including device heterogeneity, model deficiencies, rapidly increasing demands for data analysis, and limitations of cloud computing, this study proposes an Internet of Things (IoT) architecture. This architecture is centered around an intelligently decoupled gateway design and edge-cloud collaborative intelligent meat inspection. Pork freshness detection is used as an example. In this paper, a high-precision and lightweight pork freshness detection model is developed by optimizing the MobileNetV3 model with Efficient Channel Attention (ECA). The experimental results indicate that the model’s accuracy on the test set is 99.8%, with a loss function value of 0.019. Building upon these results, this paper presents an experimental platform for real-time pork freshness detection, implemented by deploying the model on an intelligent gateway. The platform demonstrates stable performance with peak model memory usage under 600 MB, average CPU utilization below 20%, and gateway internal response times not exceeding 100 ms.
Read full abstract