A Pedestrian Re-Identification Method Based on Erasing Salient Areas

Abstract

The invention discloses a pedestrian re-identification method based on erasing salient regions, including: (1) randomly selecting P different people from the training data, each person randomly selects K pictures, and obtains P×K pictures as One training data; (2) Obtain the salient area of ​​the picture according to the similarity between the pictures; (3) Erase a certain proportion of the salient area according to the preset erasing area and probability; (4) Extract the erasing The feature of the picture in the salient area is obtained and the adaptive global pooling layer is used to perform the pooling operation, and the error is calculated by using the feature; (5) The feature vector of the original image is calculated, and the error is calculated by using the feature vector; (6) The combination step The errors obtained in (4) and (5) are returned to the gradient training model; (7) use the trained model for pedestrian re-identification applications. Utilizing the present invention can help the model learn better and richer feature expressions and improve the re-identification effect.

Description

technical field [0001] The invention belongs to the field of pedestrian re-identification, in particular to a pedestrian re-identification method based on erasing salient regions. Background technique [0002] Person re-identification is a very important research area, given a picture of a person, it needs to return another picture of this person in the database. Due to the widespread application of deep learning, there has been rapid development and progress in this field in the past few years. However, due to occlusions, changes in pedestrian poses, background effects, and different camera perspectives, pedestrian re-identification is still a very challenging task. Pedestrian re-identification plays a very important role in the field of security. For example, it can query children separated from their parents in amusement parks, and use urban cameras to quickly locate and track criminal suspects. Therefore, how to improve the performance of re-identification has attracte...

AI Technical Summary

Problems solved by technology

However, as the training continues, the model will pay more and more attention to the visual information whose significance is easy to judge, which will cause the model to ignore other visual features and reduce its accuracy, especially when the pedestrians in the two pictures are not clearly distinguished. The effect will be greatly reduced

Images