Fisheye Person Detection via Positional Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for detecting people using fisheye camera images face challenges such as increased processing costs and delays in real-time detection due to preprocessing requirements, and suffer from erroneous detection due to distortion and object similarity, leading to decreased accuracy.
Innovation Solution
A human detection device that analyzes fisheye images without preprocessing, using a detector to identify human body candidates and objects, and an erroneous detection determination unit to assess positional relationships between candidates and objects to determine accurate human presence, including posture and motion estimation based on object interactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If planner development processing is performed to correct fisheye image distortion, then detection accuracy is improved, but processing time increases and real-time detection becomes difficult
Solution Approach 1:
The invention extracts only the essential detection information from the fisheye image without performing complete planner development. The system detects human body candidates directly from the distorted fisheye image and then applies selective correction only when needed, rather than converting the entire image beforehand. This extraction approach maintains detection accuracy while significantly reducing processing time.
Solution Approach 2:
The invention performs preliminary detection of human body candidates from the raw fisheye image before any correction processing. By identifying potential targets in advance and only applying correction to specific regions or frames where detection is required, the system achieves real-time performance while maintaining accuracy. The preliminary action of candidate identification allows subsequent correction to be performed more efficiently.
2Productivity
If fisheye image is used directly without planner development, then processing speed is improved, but detection accuracy deteriorates due to distortion and erroneous detection
Solution Approach 1:
The invention applies different processing qualities to different regions of the fisheye image. Instead of uniformly correcting the entire image, the system identifies regions containing human body candidates and applies correction locally to those specific areas. This local quality approach maintains high processing speed for the majority of the image while ensuring accurate detection in critical regions, thereby resolving the contradiction between speed and accuracy.
Solution Approach 2:
The system uses feedback from initial detection results to guide subsequent processing. By first detecting human body candidates from the raw fisheye image and then using this information to determine where correction is needed, the system continuously refines its detection accuracy. This feedback mechanism allows the system to maintain high speed while improving accuracy through iterative refinement based on detection results.
3Loss of time
If human body detection is performed on distorted fisheye image, then real-time detection is enabled, but erroneous detection increases due to object similarity and distortion
Solution Approach 1:
The invention segments the detection process into distinct stages: first identifying human body candidates from the distorted fisheye image, then separately performing verification and correction. This segmentation allows the system to maintain real-time detection by processing candidates in discrete, manageable steps rather than attempting to correct the entire image at once. The segmented approach improves reliability by allowing verification of each candidate against the original distorted image characteristics.
Solution Approach 2:
The system introduces an intermediary verification step between initial detection and final confirmation. This intermediary unit checks whether detected human body candidates are indeed accurate by comparing them against the distorted fisheye image characteristics and surrounding context. The intermediary acts as a mediator that filters out erroneous detections while maintaining real-time performance, thereby improving detection reliability without significant time penalty.
Data Source
AI summary
A human detection device, which analyzes a fisheye image obtained by a fisheye camera installed above a detection target area to detect a person present in the detection target area, includes: a detector configured to detect a human body candidate and an object from a fisheye image; and an erroneous detection determination unit configured to determine whether or not the human body candidate has been erroneously detected based on a positional relationship between a detection position of the human body candidate and a detection position of the object present around the human body candidate.