Abstract: Product data for a product is received by an attribute selection module. The product data includes product image data and product text data. This product data is used to generate a plurality of probability distributions for a category. The category includes a plurality of attributes, and the probability distribution includes a plurality of probabilities indicating the likelihoods that attributes of the category are applicable to the product. The plurality of probability distributions for the category are weighted and summed to generate a combined probability distribution for the category. An attribute label is determined by selecting an attribute from the category that is indicated to be most likely applicable to the product based on the combined probability distribution for the category. The attribute label is associated with the product. The attribute label enables other services to search for and retrieve the product based on the attribute.

Abstract: Product data for a product is received by an attribute selection module. The product data includes product image data and product text data. This product data is used to generate a plurality of probability distributions for a category. The category includes a plurality of attributes, and the probability distribution includes a plurality of probabilities indicating the likelihoods that attributes of the category are applicable to the product. The plurality of probability distributions for the category are weighted and summed to generate a combined probability distribution for the category. An attribute label is determined by selecting an attribute from the category that is indicated to be most likely applicable to the product based on the combined probability distribution for the category. The attribute label is associated with the product. The attribute label enables other services to search for and retrieve the product based on the attribute.

Abstract: A detection and classification method of a shape in a medical image is provided. It is based on generating a plurality of 2-D sections through a 3-D volume in the medical image. In general, there are two steps. The first step is feature estimation to generate shape signatures for candidate volumes containing candidate shapes. The feature estimation method computes descriptors of objects or of their images. The second general step involves classification of these shape signatures for diagnosis. A classifier contains, builds and/or trains a database of descriptors for previously seen shapes, and then maps descriptors of novel images to categories corresponding to previously seen shapes or classes of shapes.

Abstract: A method to detect and classify a structure of interest in a medical image is provided to enable high specificity without sacrificing the sensitivity of detection. The method is based on representing changes in three-dimensional image data with a vector field, characterizing the topology of this vector field and using the characterized topology of the vector field for classification of a structure of interest. The method could be used as a stand-alone method or as a post-processing method to enhance and classify outputs of a high-sensitivity low-specificity method to eliminate false positives.

Abstract: A detection and classification method of a shape in a medical image is provided. It is based on generating a plurality of 2-D sections through a 3-D volume in the medical image. In general, there are two steps. The first step is feature estimation to generate shape signatures for candidate volumes containing candidate shapes. The feature estimation method computes descriptors of objects or of their images. The second general step involves classification of these shape signatures for diagnosis. A classifier contains, builds and/or trains a database of descriptors for previously seen shapes, and then maps descriptors of novel images to categories corresponding to previously seen shapes or classes of shapes.

Abstract: A method to detect and classify a structure of interest in a medical image is provided to enable high specificity without sacrificing the sensitivity of detection. The method is based on representing changes in three-dimensional image data with a vector field, characterizing the topology of this vector field and using the characterized topology of the vector field for classification of a structure of interest. The method could be used as a stand-alone method or as a post-processing method to enhance and classify outputs of a high-sensitivity low-specificity method to eliminate false positives.