Configuring system operation using image data

摘要

A system that configures a device's operation based on the device's environment. The system may receive scene data describing a scene in which the device will operate. The scene data may include image data, audio data, or other data. A feature vector comprising the scene data may be processed to identify one or more categories to be associated with the scene. Various processing techniques, such as using Bayesian nonparametric models, may be used to categorize the scene data. The device may then adjust its operation based on the one or more selected categories.

主权项

1. A computer-implemented method comprising:
receiving image data, the image data corresponding to an operating environment of a device; receiving audio data, the audio data corresponding to the environment; constructing a first feature vector using at least the image data and the audio data; processing the first feature vector using a Bayesian nonparametric technique to associate an environment category to the environment, the technique comprising:
comparing the first feature vector to a reference feature vector, the reference feature vector corresponding to a first environment category,determining an association likelihood based on the comparing, the association likelihood indicating a likelihood that the first environment category describes the operating environment of the device,determining a prior score for the first environment category, the prior score based on an initial bias, wherein the initial bias is determined using a number of times the first environment category has been selected previously,determining an adjusted score for the first environment category using the association likelihood and the prior score,determining a new category score representing a likelihood that the first feature vector does not correspond to any established environment category, the new category score based on the initial bias,comparing the adjusted score to the new category score, andselecting the first environment category as corresponding to the environment based on the adjusted score being higher than the new category score; and setting a speech processing parameter of the device based on the selected first environment category.