A RFID-based Diet Estimation of Grower Pigs

In the study, embedded devices combined with wireless communication technology, Radio Frequency Identification (RFIDs), are used to collect pig's diet data in the fodder area of the pigpen and transmit data to the server via WiFi/4G networks. The server is responsible for the pre-processing and feature extraction of the data. Bayes classifier is designed to derive the likelihood ratio and to infer their appetite, and the results are verified by comparing the images recorded in the pigpen. The sensitivity of the verification results was 71.1%, the singularity was 87.1%, the accuracy was 88.8%, and the precision was 26.1%. Finally, to make it easy to observe for pig farmers, the study demonstrates the numeric results of the daily meal number and the total daily feeding duration by Bout Criterion Interval (BCI). Thus, pig farmers can remotely monitor the diet of pigs and reduce the labor cost significantly.