Object detection method based on human-object interaction weak supervision label

Abstract

The invention belongs to the field of computer vision and robot vision, particularly relates to an object detection method based on a human-object interaction weak supervision label, and aims at solving the problems that an existing weak supervision object detection method is low in model convergence speed and poor in detection precision. The method comprises the following steps: acquiring a to-be-detected image as an input image; obtaining an object detection result corresponding to the input image through a trained weak supervision object detection model, wherein the weak supervision objectdetection model is constructed based on a deep convolutional neural network. According to the invention, the precision of weak supervision object detection is improved.

Description

technical field [0001] The invention belongs to the fields of computer vision and robot vision, and in particular relates to an object detection method, system and device based on human-object interaction weak supervision labels. Background technique [0002] With the popularization of mobile Internet and smart terminal devices, the amount of network image data is increasing rapidly. However, the "bounding box" label of the object required for target detection is expensive to label and the labeling process is boring, while the category-level labels in the picture are easier to obtain and the cost is low. Therefore, the weakly supervised target detection technology based on category labels is ready attention. [0003] On the one hand, most of the current weakly supervised target detection methods are designed based on the multiple instance learning (MIL) framework and analyze objects as independent individuals, such as the WSDNN model. Subsequent research has further introdu...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the above-mentioned problems in the prior art, that is, in order to solve the problem that the existing weakly supervised object detection method ignores the modeling of the interaction context between people and objects, and the modeling of the interaction relationship between people and objects is relatively simple, resulting in slow model convergence and low detection accuracy. poor problem, the present invention proposes an object detection method based on human-object interaction weakly supervised labels, the method includes:

Images