3D Coordinate Mapping for Dense Pedestrian Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection methods struggle with accurate detection in scenarios with high pedestrian densities and complex traffic conditions, often resulting in incomplete pedestrian location due to computational complexity and shielding issues.
Innovation Solution
The method involves mapping image frames into a 3D physical space to obtain pixel coordinates, extracting foreground regions, segmenting them into blobs, and using a neural network for object detection based on 3D coordinates of predetermined reference points within each blob.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If pedestrian detection is performed directly for each frame of the video, then detection coverage is improved, but computation time increases significantly and the algorithm becomes complicated
Solution Approach 1:
The video is segmented into multiple key frames at regular intervals, and pedestrian detection is performed only on these key frames rather than every single frame. This segmentation approach maintains detection coverage while significantly reducing computation time and algorithm complexity.
2Ease of operation
If blob-based pedestrian tracking is used, then tracking simplicity is improved, but detection accuracy deteriorates in high-density scenarios where each blob contains multiple pedestrians
Solution Approach 1:
The foreground region is segmented into multiple blobs based on spatial separation, and each blob is further divided into multiple regions of interest (ROIs). This multi-level segmentation ensures that even in high-density scenarios, each ROI contains at most one pedestrian, maintaining both tracking simplicity and location accuracy.
Solution Approach 2:
The patent introduces a vertical dimension by dividing each blob into multiple ROIs vertically, creating a hierarchical structure where blobs are first separated horizontally, then each blob is subdivided vertically into multiple ROIs. This dimensional division ensures accurate pedestrian localization even when pedestrians are densely packed.
3Device complexity
If traditional object detection algorithms are used in high-density scenarios, then algorithm simplicity is maintained, but detection accuracy deteriorates due to shielding situations and high pedestrian density
Solution Approach 1:
The image is segmented into multiple blobs based on foreground detection, and each blob is further divided into multiple ROIs. This segmentation reduces the search space for detection algorithms and minimizes shielding effects, improving detection accuracy without significantly increasing algorithm complexity.
Solution Approach 2:
Different regions (ROIs) within blobs are treated with different detection strategies. By dividing blobs into multiple ROIs, the algorithm can apply localized detection parameters to each region, improving detection accuracy in high-density areas while maintaining overall algorithm simplicity.
Data Source
AI summary
An object detection method and an object detection apparatus are provided. The object detection method includes: mapping at least one image frame in an image sequence into a three dimensional physical space to obtain three dimensional coordinates of each pixel in the at least one image frame; extracting a foreground region in the at least one image frame; segmenting the foreground region into a set of blobs; and detecting, for each blob in the set of blobs, an object in the blob through a neural network based on the three dimensional coordinates of at least one predetermined reference point in the blob, to obtain an object detection result.