Abstract: Techniques for automating workflow using a real-time driver activity classification model are described herein. For example, an electronic device can execute a machine learning model based at least in part on an input comprising geolocation data and motion data detected by the electronic device. The machine learning model can generate a classification indicating that a state of the electronic device with respect to motion. The electronic device can detect that the electronic device is within a threshold distance from a service address. The electronic device can detect that the state is a non-driving state. Responsive to (i) detecting that the state is a non-driving state and (ii) detecting that the electronic device is within the threshold distance from the service address, the electronic device can transition from presenting a set of driving instructions to presenting instructions for performing a service at the service address.

Abstract: Various proxies can be used to qualitatively or quantitatively assess welfare of a poultry flock. For example, as shown and described herein, one or more of a video-based monitoring technique or audio-based monitoring technique can be used. Such monitoring can include a video-based (e.g., image based) technique where indicia of movement of individual members of the flock are automatically tracked. In addition, or instead, an audio-based technique can be used, such as for classification of vocalization. Various metrics can be stored or reported using such techniques and one or more alerts can be generated, such as indicative of flock welfare or a change in flock welfare. Techniques described herein can include identifying paths traversed by respective members of a flock (e.g., individual animals), and then extracting indications such as corresponding to a distribution of such paths. Such paths can be defined by corresponding movement trajectories associated with individual members.

Abstract: Various proxies can be used to qualitatively or quantitatively assess welfare of a poultry flock. For example, as shown and described herein, one or more of a video-based monitoring technique or audio-based monitoring technique can be used. Such monitoring can include a video-based (e.g., image based) technique where indicia of movement of individual members of the flock are automatically tracked. In addition, or instead, an audio-based technique can be used, such as for classification of vocalization. Various metrics can be stored or reported using such techniques and one or more alerts can be generated, such as indicative of flock welfare or a change in flock welfare. Techniques described herein can include identifying paths traversed by respective members of a flock (e.g., individual animals), and then extracting indications such as corresponding to a distribution of such paths. Such paths can be defined by corresponding movement trajectories associated with individual members.