Development and Validation of an Automated Video Tracking Model for Stabled Horses

Analysis of Various Facial Expressions of Horses as a Welfare Indicator Using Deep Learning

This study aimed to prove that deep learning can be effectively used for identifying various equine facial expressions as welfare indicators. In this study, a total of 749 horses (healthy: 586 and experiencing pain: 163) were investigated. Moreover, a model for recognizing facial expressions based on images and their classification into four categories, i.e., resting horses (RH), horses with pain (HP), horses immediately after exercise (HE), and horseshoeing horses (HH), was developed. The normalization of equine facial posture revealed that the profile (99.45%) had higher accuracy than the front (97.59%). The eyes-nose-ears detection model achieved an accuracy of 98.75% in training, 81.44% in validation, and 88.1% in testing, with an average accuracy of 89.43%. Overall, the average classification accuracy was high; however, the accuracy of pain classification was low. These results imply that various facial expressions in addition to pain may exist in horses depending on the situation, degree of pain, and type of pain experienced by horses. Furthermore, automatic pain and stress recognition would greatly enhance the identification of pain and other emotional states, thereby improving the quality of equine welfare.

Artificial Intelligence for Lameness Detection in Horses-A Preliminary Study

Lameness in horses is a long-known issue influencing the welfare, as well as the use, of a horse. Nevertheless, the detection and classification of lameness mainly occurs on a subjective basis by the owner and the veterinarian. The aim of this study was the development of a lameness detection system based on pose estimation, which permits non-invasive and easily applicable gait analysis. The use of 58 reference points on easily detectable anatomical landmarks offers various possibilities for gait evaluation using a simple setup. For this study, three groups of horses were used: one training group, one analysis group of fore and hindlimb lame horses and a control group of sound horses. The first group was used to train the network; afterwards, horses with and without lameness were evaluated. The results show that forelimb lameness can be detected by visualising the trajectories of the reference points on the head and both forelimbs. In hindlimb lameness, the stifle showed promising results as a reference point, whereas the tuber coxae were deemed unsuitable as a reference point. The study presents a feasible application of pose estimation for lameness detection, but further development using a larger dataset is essential.

Assessing the utility value of Hucul horses using classification models, based on artificial neural networks

The aim of this study was to evaluate factors influencing the performance of Hucul horses and to develop a prediction model, based on artificial neural (AI) networks for predict horses’ classification, relying on their performance value assessment during the annual Hucul championships. The Feedforward multilayer artificial neural networks, learned using supervised methods and implemented in Matlab programming environment were applied. Artificial neural networks with one and two hidden layers with different numbers of neurons equipped with a tangensoidal transition function, learned using the Levenberg-Marqiuardt method, were applied for the analysis. Although results showed that 7-year-old horses had the highest number of wins, the 11-year-old horses were observed to have had the best results when accessed relative to the total number of horses for a given year. Although horses from the Hroby line had the most starts in 2009–2019, those of the Goral line had the most wins. While predicting the horses’ efficiency for the first 6 positions during the utility championship, the neural network consisting of 12 neurons in hidden layer performed the best, obtaining 69,65% efficiency. The highest horse efficiency classification was obtained for the four-layered network with 12 and 8 neurons in the hidden layers. An 81.3% efficiency was obtained while evaluating the correctness of the prediction for horses occupying positions 1 to 3. The use of AI seems to be indispensable in assessing the performance value of Hucul horses. It is necessary to determine the relation between horses’ traits and their utility value by means of trait selection methods, accompanied with expert advice. It is also advisable to conduct research using deep neural networks.

Holstein Cattle Face Re-Identification Unifying Global and Part Feature Deep Network with Attention Mechanism

In precision dairy farming, computer vision-based approaches have been widely employed to monitor the cattle conditions (e.g., the physical, physiology, health and welfare). To this end, the accurate and effective identification of individual cow is a prerequisite. In this paper, a deep learning re-identification network model, Global and Part Network (GPN), is proposed to identify individual cow face. The GPN model, with ResNet50 as backbone network to generate a pooling of feature maps, builds three branch modules (Middle branch, Global branch and Part branch) to learn more discriminative and robust feature representation from the maps. Specifically, the Middle branch and the Global branch separately extract the global features of middle dimension and high dimension from the maps, and the Part branch extracts the local features in the unified block, all of which are integrated to act as the feature representation for cow face re-identification. By performing such strategies, the GPN model not only extracts the discriminative global and local features, but also learns the subtle differences among different cow faces. To further improve the performance of the proposed framework, a Global and Part Network with Spatial Transform (GPN-ST) model is also developed to incorporate an attention mechanism module in the Part branch. Additionally, to test the efficiency of the proposed approach, a large-scale cow face dataset is constructed, which contains 130,000 images with 3000 cows under different conditions (e.g., occlusion, change of viewpoints and illumination, blur, and background clutters). The results of various contrast experiments show that the GPN outperforms the representative re-identification methods, and the improved GPN-ST model has a higher accuracy rate (up by 2.8% and 2.2% respectively) in Rank-1 and mAP, compared with the GPN model. In conclusion, using the Global and Part feature deep network with attention mechanism can effectively ameliorate the efficiency of cow face re-identification.

Characterization of tmt-opsin2 in Medaka Fish Provides Insight Into the Interplay of Light and Temperature for Behavioral Regulation

One of the big challenges in the study of animal behavior is to combine molecular-level questions of functional genetics with meaningful combinations of environmental stimuli. Light and temperature are important external cues, influencing the behaviors of organisms. Thus, understanding the combined effect of light and temperature changes on wild-type vs. genetically modified animals is a first step to understand the role of individual genes in the ability of animals to cope with changing environments. Many behavioral traits can be extrapolated from behavioral tests performed from automated motion tracking combined with machine learning. Acquired datasets, typically complex and large, can be challenging for subsequent quantitative analyses. In this study, we investigate medaka behavior of tmt-opsin2 mutants vs. corresponding wild-types under different light and temperature conditions using automated tracking combined with a convolutional neuronal network and a Hidden Markov model-based approach. The temperatures in this study can occur in summer vs. late spring/early autumn in the natural habitat of medaka fish. Under summer-like temperature, tmt-opsin2 mutants did not exhibit changes in overall locomotion, consistent with previous observations. However, detailed analyses of fish position revealed that the tmt-opsin2 mutants spent more time in central locations of the dish, possibly because of decreased anxiety. Furthermore, a clear difference in location and overall movement was obvious between the mutant and wild-types under colder conditions. These data indicate a role of tmt-opsin2 in behavioral adjustment, at least in part possibly depending on the season.

Pain assessment in horses using automatic facial expression recognition through deep learning-based modeling

The aim of this study was to develop and evaluate a machine vision algorithm to assess the pain level in horses, using an automatic computational classifier based on the Horse Grimace Scale (HGS) and trained by machine learning method. The use of the Horse Grimace Scale is dependent on a human observer, who most of the time does not have availability to evaluate the animal for long periods and must also be well trained in order to apply the evaluation system correctly. In addition, even with adequate training, the presence of an unknown person near an animal in pain can result in behavioral changes, making the evaluation more complex. As a possible solution, the automatic video-imaging system will be able to monitor pain responses in horses more accurately and in real-time, and thus allow an earlier diagnosis and more efficient treatment for the affected animals. This study is based on assessment of facial expressions of 7 horses that underwent castration, collected through a video system positioned on the top of the feeder station, capturing images at 4 distinct timepoints daily for two days before and four days after surgical castration. A labeling process was applied to build a pain facial image database and machine learning methods were used to train the computational pain classifier. The machine vision algorithm was developed through the training of a Convolutional Neural Network (CNN) that resulted in an overall accuracy of 75.8% while classifying pain on three levels: not present, moderately present, and obviously present. While classifying between two categories (pain not present and pain present) the overall accuracy reached 88.3%. Although there are some improvements to be made in order to use the system in a daily routine, the model appears promising and capable of measuring pain on images of horses automatically through facial expressions, collected from video images.