Abstract: Techniques related to the computation of gaze vectors of users of wearable devices are disclosed. A neural network may be trained through first and second training steps. The neural network may include a set of feature encoding layers and a plurality of sets of task-specific layers that each operate on an output of the set of feature encoding layers. During the first training step, a first image of a first eye may be provided to the neural network, eye segmentation data may be generated using the neural network, and the set of feature encoding layers may be trained. During the second training step, a second image of a second eye may be provided to the neural network, network output data may be generated using the neural network, and the plurality of sets of task-specific layers may be trained.

Abstract: Techniques related to the computation of gaze vectors of users of wearable devices are disclosed. A neural network may be trained through first and second training steps. The neural network may include a set of feature encoding layers and a plurality of sets of task-specific layers that each operate on an output of the set of feature encoding layers. During the first training step, a first image of a first eye may be provided to the neural network, eye segmentation data may be generated using the neural network, and the set of feature encoding layers may be trained. During the second training step, a second image of a second eye may be provided to the neural network, network output data may be generated using the neural network, and the plurality of sets of task-specific layers may be trained.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an object recognition neural network using multiple data sources. One of the methods includes receiving training data that includes a plurality of training images from a first source and images from a second source. A set of training images are obtained from the training data. For each training image in the set of training images, contrast equalization is applied to the training image to generate a modified image. The modified image is processed using the neural network to generate an object recognition output for the modified image. A loss is determined based on errors between, for each training image in the set, the object recognition output for the modified image generated from the training image and ground-truth annotation for the training image. Parameters of the neural network are updated based on the determined loss.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an object recognition neural network using multiple data sources. One of the methods includes receiving training data that includes a plurality of training images from a first source and images from a second source. A set of training images are obtained from the training data. For each training image in the set of training images, contrast equalization is applied to the training image to generate a modified image. The modified image is processed using the neural network to generate an object recognition output for the modified image. A loss is determined based on errors between, for each training image in the set, the object recognition output for the modified image generated from the training image and ground-truth annotation for the training image. Parameters of the neural network are updated based on the determined loss.

Abstract: Systems for web page or web application instrumentation. Embodiments commence upon identification of a computer-readable user interface description comprising at least some markup language conforming to a respective document object model that is codified in a computer-readable language. An injector process modifies the user interface description by inserting markup text and code into the user interface description, where the inserted code includes instrumentation code to invoke dynamic generation of instrumentation locator IDs using the hierarchical elements found in the document object model. The modified computer-readable interface description is transmitted to a user device. Log messages are emitted upon user actions taken while using the user device. The log messages comprise the instrumentation locator IDs that are formed using hierarchical elements found in the document object model.

Abstract: Techniques related to the computation of gaze vectors of users of wearable devices are disclosed. A neural network may be trained through first and second training steps. The neural network may include a set of feature encoding layers and a plurality of sets of task-specific layers that each operate on an output of the set of feature encoding layers. During the first training step, a first image of a first eye may be provided to the neural network, eye segmentation data may be generated using the neural network, and the set of feature encoding layers may be trained. During the second training step, a second image of a second eye may be provided to the neural network, network output data may be generated using the neural network, and the plurality of sets of task-specific layers may be trained.

Abstract: Systems for managing user collaboration over objects stored on a cloud-based service platform. A server in a cloud-based platform maintains a set of read/write metadata that is associated in one-to-one correspondence to stored objects that are accessible by two or more collaborators. The server does not maintain a list of peers that communicate over peer-to-peer connections, rather the server exposes a semaphore for access to the metadata that pertains to a particular one of the stored objects. The server responds to download requests from the collaborators so as to deliver executable signaling protocol computer code to the collaborators' user devices. The signaling protocol code includes semaphore access by the collaborators' user devices using an application programming interface. Two or more accesses over the same semaphore establishes a leader and at least one follower. Corresponding ephemeral peer-to-peer connections are established between the leader and the at least one follower.

Abstract: Systems for managing user collaboration over objects stored on a cloud-based service platform. A server in a cloud-based platform maintains a set of read/write metadata that is associated in one-to-one correspondence to stored objects that are accessible by two or more collaborators. The server does not maintain a list of peers that communicate over peer-to-peer connections, rather the server exposes a semaphore for access to the metadata that pertains to a particular one of the stored objects. The server responds to download requests from the collaborators so as to deliver executable signaling protocol computer code to the collaborators' user devices. The signaling protocol code includes semaphore access by the collaborators' user devices using an application programming interface. Two or more accesses over the same semaphore establishes a leader and at least one follower. Corresponding ephemeral peer-to-peer connections are established between the leader and the at least one follower.

Abstract: Systems for managing user collaboration over objects stored on a cloud-based service platform. A server in a cloud-based platform maintains a set of read/write metadata that is associated in one-to-one correspondence to stored objects that are accessible by two or more collaborators. The server does not maintain a list of peers that communicate over peer-to-peer connections, rather the server exposes a semaphore for access to the metadata that pertains to a particular one of the stored objects. The server responds to download requests from the collaborators so as to deliver executable signaling protocol computer code to the collaborators' user devices. The signaling protocol code includes semaphore access by the collaborators' user devices using an application programming interface. Two or more accesses over the same semaphore establishes a leader and at least one follower. Corresponding ephemeral peer-to-peer connections are established between the leader and the at least one follower.

Abstract: Systems for managing user collaboration over objects stored on a cloud-based service platform. A server in a cloud-based platform maintains a set of read/write metadata that is associated in one-to-one correspondence to stored objects that are accessible by two or more collaborators. The server does not maintain a list of peers that communicate over peer-to-peer connections, rather the server exposes a semaphore for access to the metadata that pertains to a particular one of the stored objects. The server responds to download requests from the collaborators so as to deliver executable signaling protocol computer code to the collaborators' user devices. The signaling protocol code includes semaphore access by the collaborators' user devices using an application programming interface. Two or more accesses over the same semaphore establishes a leader and at least one follower. Corresponding ephemeral peer-to-peer connections are established between the leader and the at least one follower.

Abstract: Systems for web page or web application instrumentation. Embodiments commence upon identification of a computer-readable user interface description comprising at least some markup language conforming to a respective document object model that is codified in a computer-readable language. An injector process modifies the user interface description by inserting markup text and code into the user interface description, where the inserted code includes instrumentation code to invoke dynamic generation of instrumentation locator IDs using the hierarchical elements found in the document object model. The modified computer-readable interface description is transmitted to a user device. Log messages are emitted upon user actions taken while using the user device. The log messages comprise the instrumentation locator IDs that are formed using hierarchical elements found in the document object model.