Target detection method, device, electronic equipment and storage medium

Abstract

The present application proposes a target detection method, device, electronic equipment and storage medium. Wherein, the target detection method includes: obtaining the first height of the image to be detected, the first height is the height of the target; obtaining the preferred height range corresponding to multiple target detection models; based on the first height of the image to be detected and multiple For the preferred height range corresponding to the target detection model, the target detection model to be used for target detection in the image to be detected is determined from the plurality of target detection models as the first target detection model. It can provide adaptive target detection without modifying the network representation capability of the target detection model, effectively expands the target detection capability, and improves the accuracy and efficiency of target detection.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, and in particular to a target detection method, device, electronic equipment and storage medium. Background technique [0002] At present, when using the target detection model to detect objects in images, there will be multi-scale situations in the images. For example, when text detection is performed on images, there will be multi-scale text lines. Too large or too small sizes will cause the target The detection accuracy and efficiency are low. In related technologies, the optimization method aimed at the large difference in the text line scale detection index usually modifies the receptive field of the network and introduces the context information around the target, but in the actual detection application, modifying the receptive field and introducing the target context information involves complicated Hyperparameter adjustments, inappropriate receptive fields and a...

AI Technical Summary

Problems solved by technology

[0002] At present, when using the target detection model to detect objects in images, there will be multi-scale situations in the images. For example, when text detection is performed on images, there will be multi-scale text lines. Too large or too small sizes will cause the target Detection accuracy and efficiency are low

In related technologies, the optimization method aimed at the large difference in the text line scale detection index usually modifies the receptive field of the network and introduces the context information around the target, but in the actual detection application, modifying the receptive field and introducing the target context information involves complicated Hyperparameter adjustments, inappropriate receptive fields and additional contextual information will damage the representation ability of the network, and the effect is not good

Images