CN-121982727-A - Container serial number identification method and electronic equipment

CN121982727ACN 121982727 ACN121982727 ACN 121982727ACN-121982727-A

Abstract

The application discloses a container serial number identification method and electronic equipment, and the container serial number identification method comprises the steps of obtaining a container image, inputting the container image into a serial number area detection model to obtain a serial number area detection result, training a deep learning neural network model by the serial number area detection model based on a container image sample marked with a serial number label, cutting the container image according to the serial number area to obtain a serial number area image when the serial number area detection result is a first detection result, inputting the serial number area image into a character detection model to obtain a first container serial number, and training the deep learning neural network model by the character detection model based on an image sample marked with a character label. The method realizes the identification of the serial number of the container based on the serial number region detection model and the character detection model, and improves the accuracy of the identification of the serial number of the container.

Inventors

PENG ZHIYUAN
Mei Boping
Xue Chenyun
BU WENPING
LUO ZHENLAN
LIU BO
LUO HONGWEI

Assignees

广东朗翼智能科技有限公司

Dates

Publication Date: 20260505
Application Date: 20251231

Claims (10)

1. A container serial number identification method, comprising: Acquiring a container image; Inputting the container image to a serial number area detection model to obtain a serial number area detection result, wherein the serial number area detection result comprises a first detection result used for representing that the container image contains a serial number area or a second detection result used for representing that the container image does not contain the serial number area, and the serial number area detection model is obtained by training a deep learning neural network model based on a container image sample marked with a serial number label; when the detection result of the serial number area is the first detection result, cutting the container image according to the serial number area to obtain a serial number area image; Inputting the serial number area image to a character detection model to obtain a first container serial number, and training the deep learning neural network model based on an image sample marked with a character label by the character detection model.
2. The container serial number identification method of claim 1, further comprising: filtering the first container serial number to obtain a second container serial number; determining the character arrangement direction of the serial number of the second container, wherein the character arrangement direction is horizontally or vertically arranged; And sequencing the characters in the second container serial number according to the character arrangement direction to obtain a third container serial number.
3. The container serial number identification method according to claim 2, wherein said determining the character arrangement direction of the second container serial number comprises: Calculating a first projection distribution variance of characters of the second container serial number in the horizontal direction and a second projection distribution variance of characters of the second container serial number in the vertical direction respectively; and determining the character arrangement direction according to the first projection distribution variance and the second projection distribution variance.
4. A container serial number identification method according to claim 3, wherein said determining said character arrangement direction from said first projection distribution variance and said second projection distribution variance comprises: Calculating a variance difference between the first projection distribution variance and the second projection distribution variance; If the variance difference is larger than a difference threshold, determining a projection direction corresponding to the smallest distribution variance in the first projection distribution variance and the second projection distribution variance as the character arrangement direction; If the variance difference is smaller than or equal to the difference threshold, calculating the average horizontal distance and the average vertical distance between the character pairs of the second container serial number; and determining the direction corresponding to the maximum distance in the average horizontal distance and the average vertical distance as the character arrangement direction.
5. The method for identifying a serial number of a container according to claim 4, wherein determining a projection direction corresponding to a smallest distribution variance among the first projection distribution variance and the second projection distribution variance as the character arrangement direction if the variance difference is greater than a difference threshold value comprises: If the variance difference of the first projection distribution variance minus the second projection distribution variance is greater than the difference threshold, determining the character arrangement direction as the vertical arrangement; And if the variance difference of the second projection distribution variance minus the first projection distribution variance is larger than the difference threshold, determining the character arrangement direction as the horizontal arrangement.
6. The container serial number identification method according to claim 4, wherein the determining the direction corresponding to the largest distance of the average horizontal distance and the average vertical distance as the character arrangement direction includes: If the average horizontal distance is greater than the average vertical distance, determining that the character arrangement direction is the horizontal arrangement; And if the average vertical distance is larger than the average horizontal distance, determining the character arrangement direction as the vertical arrangement.
7. The method of claim 4, wherein calculating an average horizontal distance and an average vertical distance between pairs of characters of the second container serial number if the variance difference is less than or equal to the difference threshold comprises: If the variance difference is smaller than or equal to the difference threshold, calculating a horizontal distance according to the abscissa of the neighbor character pair of the second container serial number to obtain a plurality of horizontal distances; Calculating the average value of the horizontal distances to obtain the average horizontal distance; Calculating a vertical distance according to the ordinate of any character pair of the second container serial number to obtain a plurality of vertical distances; And calculating an average value of the plurality of vertical distances to obtain the average vertical distance.
8. The container serial number identification method according to any one of claims 2 to 7, characterized by further comprising: Checking the serial number of the third container to obtain a checking result; And when the verification result is a verification correct result, sending a verification record to a client associated with a rechecking person, wherein the verification record comprises the third container serial number and the verification correct result.
9. The container serial number identification method of claim 8, further comprising: when the verification result is a verification failure result, determining characters to be corrected in the third container serial number based on a preset confusable character set; Correcting the character to be corrected according to the preset confusing character set to obtain a fourth container serial number; Updating the fourth container serial number to the third container serial number, and executing the step and the subsequent steps of checking the third container serial number to obtain a checking result until the third container serial number is sent to the client or the checking times reach the preset times when the checking result is the correct checking result.
10. An electronic device, comprising: A memory; One or more processors coupled with the memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, the one or more applications configured to perform the container serial number identification method of any one of claims 1 to 9.

Description

Container serial number identification method and electronic equipment Technical Field The application belongs to the technical field of artificial intelligence, and particularly relates to a container serial number identification method and electronic equipment. Background As global trade volume continues to increase, port container throughput increases exponentially. The quick and accurate identification of the unique identification serial number of the container is a core link of port logistics automatic management, and directly influences the efficiency of cargo tracking, loading and unloading scheduling and clearance. At present, the container serial number identification mainly depends on manual visual observation and records the container serial number, and when visual fatigue occurs to a visual inspection person, the container serial number is easy to be misplaced, so that the accuracy of the container serial number identification is lower. Disclosure of Invention In view of the above, the embodiments of the present application provide a container serial number identification method and an electronic device, so as to overcome the above problems in the prior art. In a first aspect, an embodiment of the present application provides a method for identifying a serial number of a container, including: Acquiring a container image; Inputting the container image to a serial number area detection model to obtain a serial number area detection result, wherein the serial number area detection result comprises a first detection result used for representing that the container image contains a serial number area or a second detection result used for representing that the container image does not contain the serial number area, and the serial number area detection model is obtained by training a deep learning neural network model based on a container image sample marked with a serial number label; when the detection result of the serial number area is the first detection result, cutting the container image according to the serial number area to obtain a serial number area image; Inputting the serial number area image to a character detection model to obtain a first container serial number, and training the deep learning neural network model based on an image sample marked with a character label by the character detection model. Wherein, in some optional embodiments, the container serial number identification method further comprises: filtering the first container serial number to obtain a second container serial number; determining the character arrangement direction of the serial number of the second container, wherein the character arrangement direction is horizontally or vertically arranged; And sequencing the characters in the second container serial number according to the character arrangement direction to obtain a third container serial number. Wherein in some optional embodiments, the determining the character arrangement direction of the second container serial number includes: Calculating a first projection distribution variance of characters of the second container serial number in the horizontal direction and a second projection distribution variance of characters of the second container serial number in the vertical direction respectively; and determining the character arrangement direction according to the first projection distribution variance and the second projection distribution variance. Wherein in some optional embodiments, the determining the character arrangement direction according to the first projection distribution variance and the second projection distribution variance includes: Calculating a variance difference between the first projection distribution variance and the second projection distribution variance; If the variance difference is larger than a difference threshold, determining a projection direction corresponding to the smallest distribution variance in the first projection distribution variance and the second projection distribution variance as the character arrangement direction; If the variance difference is smaller than or equal to the difference threshold, calculating the average horizontal distance and the average vertical distance between the character pairs of the second container serial number; and determining the direction corresponding to the maximum distance in the average horizontal distance and the average vertical distance as the character arrangement direction. In some optional embodiments, if the variance difference is greater than a difference threshold, determining the projection direction corresponding to the smallest distribution variance of the first projection distribution variance and the second projection distribution variance as the character arrangement direction includes: If the variance difference of the first projection distribution variance minus the second projection distribution variance is greater than the difference threshold, determining the