Assessing machine learning classifiers for the detection of animals’ behavior using depth-based tracking

Abstract

There is growing interest in the automatic detection of animals' behaviors and body postures within the field of Animal Computer Interaction, and the benefits this could bring to animal welfare, enabling remote communication, welfare assessment, detection of behavioral patterns, interactive and adaptive systems, etc. Most of the works on animals' behavior recognition rely on wearable sensors to gather information about the animals' postures and movements, which are then processed using machine learning techniques. However, non-wearable mechanisms such as depth-based tracking could also make use of machine learning techniques and classifiers for the automatic detection of animals' behavior. These systems also offer the advantage of working in setups in which wearable devices would be difficult to use. This paper presents a depth-based tracking system for the automatic detection of animals' postures and body parts, as well as an exhaustive evaluation on the performance of several classification algorithms based on both a supervised and a knowledge-based approach. The evaluation of the depth-based tracking system and the different classifiers shows that the system proposed is promising for advancing the research on animals' behavior recognition within and outside the field of Animal Computer Interaction. Highlights: • Animal Computer Interaction studies could benefit from automatic behavior detection • Cats' body postures and orientation can be determined using depth-based information • Depth-based trackers offer promising accuracy rates classifying cats' postures

Seven Years after the Manifesto: Literature Review and Research Directions for Technologies in Animal Computer Interaction

“…Body, face, eye and gaze positioning have played a part in understanding human and animal behaviour in ACI through tracking gaze [112], body posture [51,114] and automated face reactions [127] similarly to HCI [128,129]. The advancements made in HCI tracking technology have not yet been fully exploited in ACI technologies, but there is an increasing corpus of ACI studies regarding animal's tracking in horses [108,109], cats [51,130] and dogs [110,112,114] (Figure 10). Williams et al [110] wanted to increase spatial accuracy for laboratory settings by using mobile head mounted, video based, eye-tracking system achieving in their work an accuracy of 2-3°.…”

“…The constraints and difficulties of tracking technologies that limit the animals' natural behaviours leave a space open within animal-computing to draw back to the original observational tracking methods in HCI to allow animals to explore technology in ordinary ways, merging early human methods with current usability methods. ACI has recently proposed image-based-humaninterpreted recognition systems with horses [108,109], orangutans [111], giraffes [105], cats [51,130] ( Figure 11) and dogs [107,114]. These non-intrusive tracking systems vary in how they operate with some using image shape recognition [51], feature and posture recognition [51,107,109,114], motion recognition [111], proximity [53], and point recognition [110,112].…”

“…The tracker used the average area of the cat's contour, number of pixels and average depth for each cluster to classify the image. The performance of this non-wearable tracker has been extensively analysed in a follow-up study [130] in which authors compared several machine learning classifiers and a greater set of features for the classification.…”

Peer Review #2 of "Identification of predictive factors of the degree of adherence to the Mediterranean diet through machine-learning techniques (v0.1)"

“…Moreover, due to the great versatility of ML techniques, they have been used in a wide variety of application areas, to discover hidden patterns in the datasets: identification and authentication of tequilas (Pérez-Caballero et al, 2017), wearable sensor data fusion (Kanjo et al, 2018), predicting the outcomes of organic reactions (Skoraczyński et al, 2017), animal behaviour detection (Pons et al, 2017) or to measure the visual complexity of images (Machado et al, 2015). In particular, ML techniques have proven to be able to uncover unimaginable relationships in very diverse fields of application, such as image or voice recognition, sentiment analysis or language translation (Li et al, 2015;Perez-de Viñaspre and Oronoz, 2015).…”

Analysis of Dogs’ Sleep Patterns Using Convolutional Neural Networks

“…While video analysis is widely applied in the context of dogs (see, e.g. [9,10]), very few works address automatic video-based analysis of dog behavior [11,12,13]. All of these works use video from 3D Kinect camera, the installation and use of which is not trivial and also quite expensive.…”