Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Monitoring chicken heat stress using deep convolutional neural networks

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  2018 ASABE Annual International Meeting  1800314.(doi:10.13031/aim.201800314)
Authors:   Chen-Yi Lin, Kuang-Wen Hsieh, Yao-Chuan Tsai, Yan-Fu Kuo
Keywords:   Poultry, heat stress, THI, Faster R-CNN, image processing

Abstract. Poultry and eggs are major dietary protein sources. Their production account for 31.19% of total volume of animal husbandry sales in Taiwan. In tropical and subtropical areas, heat stress is one of the most challenging problems for poultry industry. The detrimental effect of heat stress reduces growth rate of broilers and egg quality, even associated with sudden and massive deaths. Early detection on heat stress is a key to stabilize poultry and egg production. Conventionally, heat stress was estimated using temperature-humidity index (THI), a combination of temperature and humidity. THI is, however, an indirect indicator. The level of heat stress may vary with chicken varieties and dietary supplies to the chicken. This work proposed to monitor behaviors of broiler chicken directly using time lapsed images and deep learning algorithms. In the study, a raspberry-pi V3 and a web camera were used to acquire images of broilers at a rate of 1 fps. Convolutional neural network classifiers were developed to identify and localize the broilers in the images. The combination of chicken activity and THI values can be used as a new predicting indicator to avoid the phenomenon of chicken heat stress.

(Download PDF)    (Export to EndNotes)