Single detection method for multi-direction scene based on fully convolutional network

A fully convolutional network and detection method technology, which is applied in the field of single detection of multi-directional scene text, can solve the problems of speed reduction, improvement, and effort

CN107977620AActive Publication Date: 2018-05-01HUAZHONG UNIV OF SCI & TECH
4 Cites 65 Cited by

Patent Information

Authority / Receiving Office
CN · China
Current Assignee / Owner
Publication Date
2018-05-01

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention discloses a single detection method for a multi-direction scene based on the fully convolutional network. A fully convolutional single detection network model is constructed, end-to-endtraining can be carried out via single network needless of multi-step processing, a multi-scale feature extraction layer is combined with a text box prediction layer to detect multi-direction naturalscene characters in different sizes, length-width ratios and resolutions, a polygonal enclosure box is combined with characters to introduce less background interference, and a final text detection result can be obtained via simple non-maximal-value inhibition operations. Compared with the prior art, the detection method is simple and effective in structure, improves the accuracy, detection speedand robustness, and is high in practical application value.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention belongs to the technical field of computer vision, and more specifically relates to a single detection method for multi-directional scene text based on a fully convolutional network. Background technique

[0002] In natural scenes, scene text is one of the most common visual objects, such as road signs, license plates, text on product packaging bags, etc. Reading scene text has very useful application value, such as image-based geolocation, commodity image retrieval, traffic monitoring system, etc. Reading text in natural scenes is still challenging due to large background changes, arbitrary foreground text orientation, variable aspect ratio, and uncontrollable lighting conditions.

[0003] Traditional text detection methods usually involve multi-level processing steps, such as: generation of character or word candidate boxes, filtering of candidate boxes, combination and connection of candidate boxes. These methods require laborious ad...

Examples

Embodiment Construction

[0060] In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not constitute a conflict with each other.

[0061] Below at first explain and illustrate with regard to the technical terms of the present invention:

[0062] VGG-16: VGG-16 is a mature convolutional neural network structure proposed by ILSVRC in 2014. It contains 16 convolutional layers and fully connected layers. The structure is uniform and stable. Only 3x3 convolution and 2x2 pooling are performed from the beginning to the end. layer...