Depth-Based Object Verification for False Detection Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image analysis methods for autonomously maneuvering vehicles and robots are prone to false object detections, which can lead to unexpected maneuvers and accidents due to misleading visual cues.
Innovation Solution
A method that verifies the presence of objects in images using depth information from sensors like monocular estimation, stereo cameras, or electromagnetic interrogation, confirming the presence of objects by expected depth changes, and filtering out false detections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If object detection is performed using visual cues from images, then objects can be identified for collision avoidance, but false detections occur causing inappropriate maneuvers
Solution Approach 1:
The patent introduces depth information as an intermediary verification layer between visual object detection and maneuver execution. The depth changes detected by sensors (lidar, radar, or depth estimation from images) serve as a mediator to confirm or reject object detections from visual cues, filtering out false detections before they trigger inappropriate maneuvers
Solution Approach 2:
The system implements feedback by comparing expected depth changes (based on detected object position and size) with actual depth measurements. This feedback loop validates object detections and provides confidence scores, allowing the system to distinguish true objects from false detections and adjust maneuvers accordingly
2Reliability
If multiple sensing modalities are used to verify object presence, then false detections are reduced, but system complexity increases
Solution Approach 1:
The patent applies multi-functionality by enabling existing sensors to serve multiple purposes: cameras perform both standard object detection and depth estimation, while lidars or radars provide both primary depth measurement and verification of visual detections. This universal use of sensors reduces the need for dedicated hardware for each function
Solution Approach 2:
The system creates a depth map as a copied representation of the three-dimensional scene from two-dimensional images or separate depth sensors. This depth map copy is then used to verify object detections without requiring direct physical interaction with the scene, enabling validation while maintaining system modularity
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Reduces false object detections by validating object presence through depth information, ensuring accurate object recognition and reducing inappropriate system responses.
Implementation Method 1
depth information obtained by measuring a distance to the scenery using a beam of electromagnetic interrogation radiation
Data Source
Figure 1
Figure 2~2(c)
Figure 3~3(c)
AI summary
A method (100) for verifying the presence of objects (3) detected in a scenery (1) based on at least one image (2) of this scenery (1), comprising the steps of: • assigning (110), by a given object detector (4), at least one image region (2a) in the image (2) to an object (3) whose presence in the scenery (1) the image region (2a) indicates; • obtaining (120) depth information (5) relating to this image region (2a), said depth information (5) being indicative of a distance between a sensor that was used to acquire the image (2) and a scenery region (1a) in the scenery (1) that corresponds to the image region (2a); • determining (130) whether the depth information (5) comprises depth changes (5a*) that are to be expected given the presence of the object (3) assigned to the image region (2a) by the object detector (4); and • if this determination is positive, determining (140) that the assignment of the image region (2a) to the object (3) by the object detector (4) is a valid detection (3*) of the object (3).