Object Cluster Confidence Scoring for Autonomous Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object detection systems for autonomous vehicles face challenges in accurately determining object presence and confidence, leading to false or missed detections due to limitations in coverage value interpretation and training approaches.
Innovation Solution
The proposed solution involves a Machine Learning Model (MLM) that generates a confidence score directly interpretable as a probability, accounting for complex relationships between detected objects. Additionally, the approach allows for multiple spatial element regions to be assigned to a set of bounding box coordinates and uses soft coverage values for training, improving object detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the maximum coverage value of a cluster is used as the confidence value, then the confidence value corresponds to a specific detection, but false detections or positives occur because it only corresponds to a small portion of the image
Solution Approach 1:
The patent combines multiple coverage values from different spatial element regions into a single aggregated confidence value. Instead of using only the maximum coverage value from one region, the system merges coverage values from multiple regions that detect the same object, producing a more reliable confidence measure that reflects the object's presence across the entire image rather than just a small portion.
2Reliability
If the sum of the coverage values of a cluster is used as the confidence value, then all detections are aggregated, but the sum is unbounded and cannot be directly interpreted as a confidence or probability measure
Solution Approach 1:
The patent transforms the unbounded sum of coverage values into a bounded confidence value that can be directly interpreted as a probability measure. By applying parameter transformation, the system converts the raw sum into a normalized confidence score within a standard range (e.g., 0 to 1), enabling direct interpretation while preserving the aggregated information from multiple detections.
3Ease of manufacture
If each set of bounding box coordinates is assigned to a single spatial element region, then the training is simplified, but false detections occur when objects are close to each other or occupy multiple spatial element regions
Solution Approach 1:
The patent segments the object detection task by allowing bounding box coordinates to be assigned to multiple spatial element regions simultaneously. Instead of forcing a single-region assignment, the system divides the detection space into multiple overlapping regions, enabling objects that are close together or span multiple regions to be detected in each relevant region, thereby reducing false detections while maintaining training feasibility.
Data Source
AI summary
In various examples, detected object data representative of locations of detected objects in a field of view may be determined. One or more clusters of the detected objects may be generated based at least in part on the locations and features of the cluster may be determined for use as inputs to a machine learning model(s). A confidence score, computed by the machine learning model(s) based at least in part on the inputs, may be received, where the confidence score may be representative of a probability that the cluster corresponds to an object depicted at least partially in the field of view. Further examples provide approaches for determining ground truth data for training object detectors, such as for determining coverage values for ground truth objects using associated shapes, and for determining soft coverage values for ground truth objects.