4. Discussion
High variability in the feeding patterns of individual, healthy
pigs was found (Table 1) and is in agreement with the observations
of other researchers (Brown-Brandl et al., 2013; Hessel and Van
den Weghe, 2011).
For the HF RFID system, using two tags per pig, a sensitivity of
88.58% and a specificity of 98.34% could be achieved for a time
window of 9 s. These performance values were based on a minimal
sensitivity of 85%, but with the ROC curve in Fig. 8 the reader can analyse the sensitivity and specificity of the system using different
time windows.
The achieved values illustrate that the HF RFID system is capable
of accurately registering feeding pigs at an industrial-commercial
feeder in a commercial-like pig barn. These results correspond
to those found in literature, although the validation methods used
were different. Reiners et al. (2009) reported an identification rate
of 97.3% for a similar HF RFID system with the antenna incorporated
in the trough and no multiplexer. This identification rate corresponds
to the percentage of trough visits detected by both RFID
and video measurements. They also reported that in 64% of the
cases, the piglets were identified with a slight time delay between
RFID and video. Brown-Brandl et al. (2013) developed and validated
an LF RFID system for monitoring feeding behaviour of cattle
and swine and mention that the cattle system was in agreement
with the video 98.3% of the time, and the swine system was in
agreement 98.7% of the time. This agreement corresponds to the
percentage of 20 s (swine system) or 30 s (cattle system) intervals
where video and RFID system were identical (either both feeding
or both not feeding).
4. DiscussionHigh variability in the feeding patterns of individual, healthypigs was found (Table 1) and is in agreement with the observationsof other researchers (Brown-Brandl et al., 2013; Hessel and Vanden Weghe, 2011).For the HF RFID system, using two tags per pig, a sensitivity of88.58% and a specificity of 98.34% could be achieved for a timewindow of 9 s. These performance values were based on a minimalsensitivity of 85%, but with the ROC curve in Fig. 8 the reader can analyse the sensitivity and specificity of the system using differenttime windows.The achieved values illustrate that the HF RFID system is capableof accurately registering feeding pigs at an industrial-commercialfeeder in a commercial-like pig barn. These results correspondto those found in literature, although the validation methods usedwere different. Reiners et al. (2009) reported an identification rateof 97.3% for a similar HF RFID system with the antenna incorporatedin the trough and no multiplexer. This identification rate correspondsto the percentage of trough visits detected by both RFIDand video measurements. They also reported that in 64% of thecases, the piglets were identified with a slight time delay betweenRFID and video. Brown-Brandl et al. (2013) developed and validatedan LF RFID system for monitoring feeding behaviour of cattleand swine and mention that the cattle system was in agreementwith the video 98.3% of the time, and the swine system was inตกลง 98.7% ของเวลา ข้อตกลงนี้สอดคล้องกับการเปอร์เซ็นต์ 20 s (ระบบหมู) หรือช่วง 30 s (ระบบปศุสัตว์)ที่วิดีโอและระบบ RFID อยู่เหมือนกัน (อย่างใดอย่างหนึ่งทั้งอาหารหรือทั้งสองไม่ให้อาหาร)
การแปล กรุณารอสักครู่..