| 摘要 |
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for adjusting confidence scores of image labels for images. In one aspect, a method includes accessing images stored in an image data store, the images being associated with respective sets of labels, the labels describing content depicted in the image and having a respective confidence score that is a measure of confidence that the label accurately describes the content depicted in the image; selecting a first image from the images and determining for each of the other images and independent of the labels, a proximity score that is a measure of a relatedness of the other image to the first image; and adjusting the set of labels associated with the first image based on the respective proximity scores of the other images and the confidence scores of the labels of the other images. |
| 主权项 |
1. A method performed by a data processing apparatus, the method comprising:
accessing, by the data processing apparatus, images stored in an image data store, the images being associated with respective sets of labels, each label describing content depicted in an image and having a respective confidence score that is a measure of confidence that the label accurately describes the content depicted in the image; accessing, by the data processing apparatus, one or more proximity metric values associated with a first image of the images; accessing, by the data processing apparatus, one or more proximity metric values associated with a second image of the images; determining, by the data processing apparatus, a proximity score for the second image based on a comparison of the proximity metric values of the first image and corresponding proximity metric values of the second image, the proximity score being a measure of a relatedness of the second image to the first image; identifying, by the data processing apparatus, a second label associated with the second image, the second label having a second confidence score that is a measure of confidence that the second label accurately describes the content depicted in the second image; and generating, by the data processing apparatus, a first label for the first image based on the second label, the proximity score for the second image, and the second confidence score for the second label; and determining, for the first label, a first confidence score based on the proximity score for the second image and the second confidence score for the second label, the first confidence score being a measure of confidence that the first label accurately describes the content depicted in the first image. |
