Boe Technology Group Co., Ltd.

Beijing Boe Technology Development Co., Ltd.

Zhanfu An

Xiaojun Tang

Tingting Wang

Guangwei Huang

Rui Zheng

A pedestrian attribute recognition method based on a pedestrian attribute recognition system is provided, and the system includes at least one attribute localization module. Each attribute localization module corresponds to a plurality of pedestrian attributes; and the attribute localization module includes a spatial transformation unit and an attribute recognition unit. The method includes: extracting, by the spatial transformation unit, feature information in a discriminable region from feature information input into the spatial transformation unit, the discriminable region being related to the plurality of pedestrian attributes corresponding to the attribute localization module; and outputting, by the attribute recognition unit, recognition results of the plurality of pedestrian attributes corresponding to the attribute localization module according to the feature information in the discriminable region.

Pedestrian attribute recognition method based on a pedestrian attribute recognition system and method for training the same

The present application disclose a neural network of predicting image definition, a training method and a prediction method. The training method includes: obtaining an image set and definition labels of some images in the image set, thereby obtaining image samples with the definition labels and to-be-expanded images except for the image samples; and extracting definition features of at least some images in the image set, obtaining definition labels of at least some images in the to-be-expanded images according to the extracted definition features, correcting the definition labels of the at least some images in the to-be-expanded images to expand the image samples, and using the image samples to train the neural network of predicting image definition, thereby obtaining a trained neural network.

Neural network of predicting image definition, training method and prediction method

A method for training an object detection model includes: firstly obtaining M sample image sets; then, obtaining an initial object detection model; and finally, training the initial object detection model by using the M sample image sets to obtain the object detection model. A sample image set includes at least one sample image and object type(s) of object(s) in each sample image. An object type corresponds to one sample image set, and the M sample image sets correspond to N object types.

Growing community of inventors

Beijing, China

Zhanfu An