Image detection method and device based on sparse point cloud, equipment and storage medium

By obtaining non-zero coordinates and offsets in sparse point cloud data and dynamically adjusting the convolution kernel, the efficiency and accuracy problems of traditional convolutional neural networks when processing sparse point cloud data are solved, and more efficient object detection is achieved.

CN119693454AActive Publication Date: 2025-03-25BEIJING BAIDU NETCOM SCI & TECH CO LTD
7 Cites 0 Cited by

Patent Information

Application Number
CN202411722546.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-11-27
Publication Date
2025-03-25
Estimated Expiration
2044-11-27

AI Technical Summary

Technical Problem

When processing sparse point cloud data, traditional convolutional neural networks face the problems of large computing volume, high memory usage and low convolution kernel utilization, and it is difficult to effectively process sparse point cloud data.

Method used

By obtaining the non-zero coordinate positions in sparse point cloud data, the offset of pixel points in the image area of ​​the preset size is determined, and target detection is performed based on this information, and the shape and position of the convolution kernel are dynamically adjusted to adapt to the structure of irregular point cloud data.

Benefits of technology

The processing accuracy and object detection accuracy of sparse point cloud data are improved, the calculation amount is reduced, and the feature extraction ability and object detection efficiency are improved.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention provides an image detection method, device and equipment based on sparse point cloud and a storage medium, relates to the technical field of artificial intelligence, in particular to the technical fields of computer vision, deep learning and the like, and can be applied to scenes of automatic driving, robot navigation and the like. According to the specific implementation scheme, an image to be processed is acquired, and the coordinate position of a non-zero point in the image to be processed is determined; the to-be-processed image is sparse point cloud data, and the non-zero point represents pixel points containing pixel information in the to-be-processed image; according to the coordinate position of the non-zero point, determining the offset of a pixel point in an image area with a preset size; the offset represents the offset degree of the pixel point out of an image area with a preset size, and the non-zero points are in one-to-one correspondence with the image area; and performing target detection on an object in the to-be-processed image according to the coordinate position of each non-zero point and the offset of the pixel points in each image area. By determining the offset of the pixel points, the image processing precision is improved.
Need to check novelty before this filing date? Find Prior Art