Gait analysis of horses for lameness detection with radar sensors

Abstract

This paper presents the preliminary investigation of the use of radar signatures to detect and assess lameness of horses and its severity. Radar sensors in this context can provide attractive contactless sensing capabilities, as a complementary or alternative technology to the current techniques for lameness assessment using video-graphics and inertial sensors attached to the horses' body. The paper presents several examples of experimental data collected at the Weipers Centre Equine Hospital at the University of Glasgow, showing the micro-Doppler signatures of horses and initial results of their analysis.

Radar for assisted living in the context of Internet of Things for Health and beyond

“…Furthermore, IoT can be extended to animal welfare applications where the dairy industry, farm animals (sheep, cattle, pigs) and horses (Thoroughbreds and leisure) can benefit for lameness assessment [33], [34] and connected farms with IoT will improve significantly productivity and animal monitoring for better yield for our growing needs.…”

“…A review classifiers for activity classification [16] advise to use multiple sensors to enhance classification accuracy by covering multiple aspect angles and combat occlusions. Another way to improve accuracy is to fuse data and select the most salient features [12,18,[33][34]. Many classifiers are used in activity classification of which SVM is the most common [35].…”

Animal Lameness Detection With Radar Sensing

“…However, there is very limited work on radar for lameness detection of animals, to the best of our knowledge (with the exception of a few papers where the signature of quadrupeds is treated as a potential "confuser" for human detection [13][14]). In this paper, we expand the preliminary results on our previous work [15] by providing an initial validation of the use of radar sensing to detect lameness in dairy cows, sheep, and horses. Experimental data were collected at the facilities of the Veterinary School at the University of Glasgow, and analyzed with techniques inspired from radar automatic target recognition (microDoppler signatures, feature extraction and supervised learning for classification).…”

Activity recognition with cooperative radar systems at C and K band

“…Previous applications of these sets of features include measuring lameness in horses [18] and activity classification [13].…”

“…Each binary pair (i.e. A1 and A2 then A1 and A3 and so on) assigned positive for true class and negative for the false class [18] with a selectable binary loss function which evaluates the posterior class probability, giving a value of confidence with the decision. Ultimately, the class with the least loss value is output as the predicted class.…”