Abstract: Various embodiments are disclosed for simultaneous object localization and attribute classification using multitask deep neural networks. In an embodiment, a method comprises: obtaining, by a processing circuit, an image from an image capture device in an environment, the image including a target object in the environment; generating, by the processing circuit, predictions from the image for the target object using a multitask deep neural network, the multitask deep neural network including a network trunk and side branches, the network trunk configured for multi-scale feature extraction guided by supervision information provided by the side branches during training of the multitask deep neural network, the side branches configured as learning task-specific classifiers; and using, by the processing circuit, the predictions to localize the target object in the environment and to classify the target object and at least one attribute of the target object.

Abstract: Systems and methods for simulating behavior of agents in a virtual environment are described. Observed customer behavior data for a retail space may be received. A model for customer behavior in the retail space is generated, the model including probabilities for customer responses to a plurality of stimuli. A virtual customer agent is built based on the model for customer behavior. A virtual retail space is built and modified, the modified virtual retail space including at least one different design feature from the virtual retail space. In the modified virtual retail space, the virtual customer agent is simulated performing a predetermined list of tasks based on the probabilities for the customer responses. Simulated customer behavior data, including responses of the virtual customer agent to the modified virtual retail space, may be generated based on the simulation.

Abstract: Various embodiments are disclosed for detecting, tracking and counting objects of interest in video frames using fusion of image and depth modalities. In an embodiment, a method comprises: obtaining multiple frames of stereo image pairs from an image capturing device; rectifying each frame of the stereo image pairs; computing a stereo disparity for each frame of the stereo image pairs; determining a first set of object detections in each frame using the computed stereo disparity; determining a second set of object detections in each left or right frame of the stereo image pair using one or more machine learning models; fusing the first and second sets of object detections; and creating tracks based on the fused object detections.

Abstract: Various embodiments are disclosed for detecting, tracking and counting objects of interest in video. In an embodiment, a method of detecting and tracking objects of interest comprises: obtaining, by a computing device, multiple frames of images from an image capturing device; detecting, by the computing device, objects of interest in each frame; accumulating, by the computing device, multiple frames of object detections; creating, by the computing device, object tracks based on a batch of object detections over multiple frames; and associating, by the computing device, the object tracks over consecutive batches.

Abstract: Various embodiments are disclosed for detecting, tracking and counting objects of interest in video. In an embodiment, a method of detecting and tracking objects of interest comprises: obtaining, by a computing device, multiple frames of images from an image capturing device; detecting, by the computing device, objects of interest in each frame; accumulating, by the computing device, multiple frames of object detections; creating, by the computing device, object tracks based on a batch of object detections over multiple frames; and associating, by the computing device, the object tracks over consecutive batches.

Abstract: A space is modeled using images from a set of cameras in the space. An easy to use tool is provided that allows users to identify a reference location in an image and a corresponding reference location on a floor plan of the space. Based on these correspondences, a model is generated that can map a point in a camera's view to a point on the floor plan, or vice-versa. Subjects moving through the space are identified and tracked using the model and attributes associated with a subject. In a specific implementation, the attributes include velocity, color, size, and position. In a specific implementation, the space is a retail store and the subjects are customers browsing through the retail store.

Abstract: Various embodiments are disclosed for detecting, tracking and counting objects of interest in video. In an embodiment, a method of detecting and tracking objects of interest comprises: obtaining, by a computing device, multiple frames of images from an image capturing device; detecting, by the computing device, objects of interest in each frame; accumulating, by the computing device, multiple frames of object detections; creating, by the computing device, object tracks based on a batch of object detections over multiple frames; and associating, by the computing device, the object tracks over consecutive batches.

Abstract: A wireless analytics system and method are disclosed that take shoppers through an initial registration process, after which they can access the Internet quickly and easily every time they visit any of the retailer's stores. During future visits to any of the retailer's store locations the shopper's device is automatically granted network access as soon as it is within wireless range. Once on the network, a shopper can access websites, social networks, and shopping resources from across the mobile web. The system collects information about these access events, including the location and timestamp of each access event. The retailer may be notified when a shopper enters their store. This information may automate the check-in process for processes within the retail store or within a mobile application.

Abstract: Systems and methods are described for excluding individuals from retail analytics. Pedestrian traffic within a predetermined area is monitored via a video sensor to determine a video entrance. The video entrance may correspond to entrances of the predetermined area by a first set of individuals. The pedestrian traffic may also be monitored via a plurality of wireless sensors to determine a wireless entrance. The wireless entrance may correspond to entrances by a second set of individuals. The video entrance and the wireless entrance are compared to determine if any of the entrances associated with a video timestamp match any of the entrances with a wireless timestamp. An overall entrance count corresponding to the number of customers entering the predetermined area is determined. Retail analytics excluding employees may subsequently be calculated using the overall entrance count.

Abstract: In an implementation, techniques are provided for tracking the location of a device that transmits wireless signal. The wireless signal can be received at a set of receivers. Distances to the device are determined based on signal strength. From the distances, boundaries encompassing possible locations of the device are defined. Boundary intersections are analyzed in order to determine the location of the device.

Abstract: In some implementations, video camera networks can be used to track people or objects and determine their locations. In some implementations, location accuracy better than one meter can be achieved using video cameras. According to some implementations, Wi-Fi localization and video localization systems can be fused together to perform calibration. In some implementations, by using the video network to continuously update the Wi-Fi localization calibration database, Wi-Fi location accuracy can be improved.

Abstract: Techniques are provided for tracking different types of subjects and labeling the tracks according to subject type. In an implementation, the tracking includes tracking first and second subject types using video, and also tracking subjects of the first type using Wi-Fi tags provided to the subjects of the first type. The video and Wi-Fi tracks can be compared in order to identify and label which video tracks are associated with subjects of the first type and which video tracks are associated with subjects of the second type. Upon the tracks having been labeled, interactions between the different subject types can be identified and analyzed.

Abstract: According to some implementations, an estimate of a target's location can be calculated by correlating Wi-Fi and video location measurements. This spatio-temporal correlation combines the Wi-Fi and video measurements to determine an identity and location of an object. The accuracy of the video localization and the identity from the Wi-Fi network provide an accurate location of the Wi-Fi identified object.

Abstract: In an implementation, techniques are provided for tracking the location of a device that transmits wireless signal. The wireless signal can be received at a set of receivers. Distances to the device are determined based on signal strength. From the distances, boundaries encompassing possible locations of the device are defined. Boundary intersections are analyzed in order to determine the location of the device.

Abstract: An image from a camera may be received and location coordinates of customers in the image may be identified. A line may be calculated based on the location coordinates and a join zone may be defined on the line. A location on the image corresponding to the join zone may be monitored. If a customer stands in the join zone for an amount of time, then the customer may be determined to have joined the customer queue. In response to the customer joining the customer queue, a new line may be calculated based on location coordinates of the first, second, and third customers in the image. Furthermore, a new join zone may be defined on the new line and a location on the image or video corresponding to the new join zone may be monitored for a subsequent customer joining the customer queue.

Abstract: In an implementation, techniques are provided for tracking the location of a device that transmits wireless signal. The wireless signal can be received at a set of receivers. Distances to the device are determined based on signal strength. From the distances, boundaries encompassing possible locations of the device are defined. Boundary intersections are analyzed in order to determine the location of the device.