Method for sorting vertex coordinate data of target detection result

Abstract

The invention provides a method for sorting vertex coordinate data of a target detection result, which belongs to the technical field of computer vision and image processing, and comprises the following steps of: searching position information of an image by using a bounding box, and obtaining data information reflecting the depth of the bounding box according to horizontal and vertical position information of a selected position of the bounding box; finding out a depth position reflecting the bounding box in the image; then forming a comprehensive information index by using horizontal and vertical position information of the selected position of the bounding box and data information of the depth of the bounding box; the relative position relation of the bounding box in the image is reflected through the multi-dimensional information contained in the information index. The problem of inaccurate sorting is solved, and the efficiency and accuracy of image processing and computer vision tasks are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of computer vision and image processing, and in particular to a method for sorting vertex coordinate data of target detection results. Background Art

[0002] In computer vision and image processing, object detection is an important task that aims to identify objects of interest from images or videos and give their location information (such as vertex coordinates of bounding boxes). However, existing object detection algorithms often only focus on the location and classification of objects, and lack an effective sorting mechanism for the spatial relationship between multiple detected objects, making the output data disorganized and affecting the efficiency of manual data verification. Summary of the invention

[0003] In order to solve the above technical problems, the present invention provides a method for effectively sorting the detected target vertex coordinate data in the target detection task, so as to solve the problem that the results of image recognition output are inconsistent with the spatial position order of the detected targets in the image, thereby effectively improving the efficiency of manual data verification, more intuitively displaying the value of data, and enhancing the user experience.

[0004] The technical solution of the present invention is:

[0005] A method for sorting vertex coordinate data of target detection results uses a bounding box to find the position information of an image, and then obtains data information reflecting the depth of the bounding box based on the horizontal and vertical position information of the selected position of the bounding box, thereby finding the depth position of the bounding box in the image; then uses the horizontal and vertical position information of the selected position of the bounding box and the data information of the depth of the bounding box to form a comprehensive information index; and reflects the relative position relationship of the bounding box in the image through the multi-dimensional information contained in the information index.

[0006] Furthermore,

[0007] include

[0008] 1) Extract bounding box coordinates;

[0009] 2) Calculate the Euclidean distance;

[0010] 3) Weighted depth distance;

[0011] 4) Synthetic composite distance;

[0012] 5) Perform sorting;

[0013] 6) Return the sorting results.

[0014] Going further,

[0015] Extract the coordinates of the lower right corner of each bounding box; calculate the Euclidean distance from each bounding box to the image origin according to the Pythagorean theorem.

[0016] Multiply the y-coordinate value of the bounding box by the preset weight to get the depth-weighted distance.

[0017] The Euclidean distance is added to the depth-weighted distance to form a composite distance metric; the bounding box list is then sorted based on the composite distance metric.

[0018] Going further,

[0019] For each bounding box, first use the Pythagorean theorem to calculate the Euclidean distance from its lower right corner to the image origin, that is, (\sqrt{x^2+y^2}); multiply the y coordinate value by 8 (that is, (y\times 8)) to obtain an additional distance value reflecting the depth of the bounding box.

[0020] The Euclidean distance is added to the additional distance value to form a composite distance metric that combines the bounding box position and depth information.

[0021] Based on the calculated composite distance metric, all bounding boxes are ranked.

[0022] The beneficial effects of the present invention are

[0023] 1. Significantly improve sorting accuracy: By comprehensively considering the position and depth information of the bounding box, the sorting algorithm of the present invention can generate a more accurate sorting result that conforms to the intuitive visual experience, effectively avoiding the sorting deviation caused by ignoring depth information in traditional methods, and significantly improving the accuracy of subsequent image analysis and target detection.

[0024] 2. Enhanced depth perception: The introduced depth weight factor makes the sorting result reflect not only the planar position of the bounding box, but also its depth order in the image space. This is particularly important when dealing with advanced computer vision tasks such as 3D scene reconstruction and stereo vision, and greatly enhances the algorithm's depth perception ability.

[0025] 3. Optimize processing efficiency: This algorithm avoids complex and redundant operations through simple and efficient calculation steps, ensuring the rapid response of the sorting process. Especially when processing large-scale image data sets, it can significantly shorten the processing time and improve the operating efficiency of the overall system.

[0026] 4. Wide applicability and compatibility: The sorting algorithm of the present invention is flexible in design and can be easily integrated into the existing image processing framework. It can be seamlessly connected to both static image analysis and real-time video stream processing, showing strong adaptability and a wide range of applicability.

[0027] In summary, the present invention not only solves the key problems existing in the traditional bounding box sorting algorithm, but also significantly improves the accuracy and efficiency of image processing and computer vision tasks by introducing a depth-aware sorting strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a schematic diagram of the workflow of the present invention. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0030] The present invention provides a method for sorting vertex coordinate data of target detection results, comprising the following steps:

[0031] 1. Extract the coordinates of the lower right corner of each bounding box;

[0032] 2. Calculate the Euclidean distance from each bounding box to the image origin according to the Pythagorean theorem;

[0033] 3. Multiply the y coordinate value of the bounding box by the preset weight to get the depth weighted distance;

[0034] 4. Add the Euclidean distance and the depth-weighted distance to form a composite distance index;

[0035] 5. Sort the bounding box list based on the composite distance metric.

[0036] The specific solutions are as follows:

[0037] 1. Data preparation stage: From the list of bounding boxes to be sorted, extract the coordinates of the lower right corner of each bounding box, that is, the x and y coordinates. These coordinate values ​​represent the specific position of the bounding box in the image and are the basis for subsequent calculations.

[0038] 2. Calculation of composite distance index: For each bounding box, the Pythagorean theorem is first used to calculate the Euclidean distance from its lower right corner to the origin of the image, that is, (\sqrt{x^2+y^2}). This calculation step ensures that the sorting can simultaneously consider the horizontal and vertical position information of the bounding box, enhancing the multi-dimensional characteristics of the sorting. Subsequently, in order to further highlight the importance of the depth information of the bounding box, a weight factor is introduced to multiply the y coordinate value by 8 (i.e. (y\times 8)) to obtain an additional distance value reflecting the depth of the bounding box. The introduction of this additional distance value enables the sorting result to more accurately reflect the depth position of the bounding box in the image, overcoming the limitation of ignoring depth information in traditional methods. Finally, the Euclidean distance is added to the additional distance value to form a composite distance index that combines the bounding box position and depth information.

[0039] 3. Sorting execution phase: All bounding boxes are sorted based on the calculated composite distance index. This sorting process makes full use of the multi-dimensional information contained in the composite distance index, ensuring that the sorting results can not only accurately reflect the relative position relationship of the bounding boxes in the image, but also effectively distinguish bounding boxes at similar horizontal positions but different depths, thereby significantly improving the accuracy and practicality of the sorting.

[0040] The above description is only a preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention, and is not used to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A method for sorting vertex coordinate data of target detection results, It is characterized in that The position information of the image is found by using the bounding box, and then the data information reflecting the depth of the bounding box is obtained according to the horizontal and vertical position information of the selected position of the bounding box, thereby finding the depth position of the bounding box in the image; then the horizontal and vertical position information of the selected position of the bounding box and the data information of the depth of the bounding box are used to form a comprehensive information index; the relative position relationship of the bounding box in the image is reflected through the multi-dimensional information contained in the information index.

2. The method according to claim 1, It is characterized in that include 1) Extract bounding box coordinates; 2) Calculate the Euclidean distance; 3) Weighted depth distance; 4) Synthetic composite distance; 5) Perform sorting; 6) Return the sorting results.

3. The method according to claim 2, It is characterized in that Extract the coordinates of the lower right corner of each bounding box; calculate the Euclidean distance from each bounding box to the image origin according to the Pythagorean theorem.

4. The method according to claim 3, It is characterized in that Multiply the y-coordinate value of the bounding box by the preset weight to get the depth-weighted distance.

5. The method according to claim 4, It is characterized in that The Euclidean distance is added to the depth-weighted distance to form a composite distance metric; the bounding box list is then sorted based on the composite distance metric.

6. The method according to claim 5, It is characterized in that For each bounding box, first use the Pythagorean theorem to calculate the Euclidean distance from its lower right corner to the image origin, that is, (\sqrt{x^2+y^2}); multiply the y coordinate value by 8 (that is, (y\times 8)) to obtain an additional distance value reflecting the depth of the bounding box.

7. The method according to claim 6, It is characterized in that The Euclidean distance is added to the additional distance value to form a composite distance metric that combines the bounding box position and depth information.

8. The method according to claim 7, It is characterized in that Based on the calculated composite distance metric, all bounding boxes are ranked.

Images