Use Of Depth Image Analysis To Detect Inactivity And Estimate Live Weight Of Pigs

Depth image analysis has a lot of potential in agriculture,especially in animal husbandry.Weight estimation represents an essential part of the livestock breeding process.Excessive inactivity in farm animals can be an early indication of ill health.The traditional way for detecting inactivity in pigs relies on manual inspection which can be laborious and especially time-consuming.This thesis discusses the methodologies involved in the application of depth images in detecting inactivity in sows housed in a pen as well as in estimating the weight of individual piglets housed in another pen.The first part discusses the application of depth images in detecting inactivity automatically.The developed system recorded sequential depth images of the animals in a pen and implemented a proposed image processing and logic analysis scheme named as 'DepInact'to keep track of the inactive time of group-housed individual pigs over time.To verify the robustness and accuracy of the developed system,a total of 656 pairs of corresponding depth data and colour images,consecutively taken four seconds apart from each other,were captured.The verification process involved manually identifying all pigs using the colour images captured.The results of identification of all pigs that were inactive for more than the preset period of time by DepInact were compared to those by manual inspection through the colour images captured.An accuracy of 85.7%was achieved using the verification data,thus demonstrating that the developed system is a viable alternative to manual detection of inactivity of group-housed pigs.The system described here is insensitive to changing light conditions in the pen house since it makes use of depth data.The second part of this research discusses the image processing sequence applied to automatically extract an individual piglets' features like back area,height and height summation from a depth image and how these features were used to estimate the piglets'weight.An SVM model was used to estimate the weight.From the model,the result of using only one input feature as well as the result from using three input features(back area,height and height summation)were recorded.The best result was obtained when the three input features were used(R=0.82,RMSE=0.09 for the calibration set,R=0.95,RMSE=0.08 for the prediction set).Nevertheless,it was suggested more research is still needed to improve the accuracy of the developed system.