Neural Network Occluded Object Detection via Visual Indicators
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous vehicle perception systems struggle to accurately detect occluded objects in environments, leading to potential safety concerns due to wrongful predictions and high computing resource requirements.
Innovation Solution
A neural network is trained using real-world and simulated image data to identify occluded objects by recognizing indicators such as shadows, reflections, and light emitted by occluded objects, allowing for more accurate detection without excessive computational resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional systems predict occluded objects in all occluded areas, then safety is improved, but computing resources are excessively consumed and wrongful predictions increase
Solution Approach 1:
The patent applies local quality by training the neural network to focus on specific local indicators (shadows, reflections, light emissions) rather than uniformly analyzing all occluded areas. This selective approach concentrates computational resources on detecting actual occluded objects through their visual indicators, reducing overall computing resource consumption while maintaining safety.
Solution Approach 2:
The patent utilizes optical property changes (shadows, reflections, light emissions) as indicators of occluded objects. By detecting these visual changes and anomalies in the environment, the system can identify occluded objects without requiring exhaustive analysis of all occluded areas, thereby reducing computational complexity while improving detection accuracy.
2Reliability
If conventional systems predict occluded objects in all occluded areas, then safety is improved, but wrongful predictions increase
Solution Approach 1:
The patent detects occluded objects by identifying specific visual indicators such as shadows, reflections, and light emissions. These optical property changes serve as reliable signatures of occluded objects, enabling the system to make accurate predictions without wrongful detections, thus improving measurement precision while maintaining safety.
Solution Approach 2:
The patent replaces conventional mechanical or rule-based detection systems with a neural network that learns to recognize patterns of occluded objects through training data. This substitution enables the system to distinguish between actual occluded objects and false indicators, significantly reducing wrongful predictions while maintaining high safety standards.
3Reliability
If conventional systems use reactions of other vehicles to predict occluded objects, then some safety improvement is achieved, but wrongful predictions still occur
Solution Approach 1:
The patent replaces behavioral inference methods (analyzing vehicle reactions) with direct visual detection using a neural network. By training the network to recognize visual indicators of occluded objects such as shadows and reflections, the system achieves more accurate and reliable detection without the wrongful predictions that arise from inferring object presence from vehicle reactions.
Solution Approach 2:
The patent utilizes optical property changes (shadows, reflections, light emissions) as direct indicators of occluded objects, providing more reliable and accurate detection compared to inferring object presence from vehicle reactions. This approach reduces wrongful predictions by focusing on physical evidence rather than behavioral inference.
Data Source
AI summary
In various examples, techniques for detecting occluded objects within an environment are described. For instance, systems and methods may receive training data representing images and ground truth data indicating whether the images are associated with occluded objects or whether the images are not associated with occluded objects. The systems and methods may then train a neural network to detect occluded objects using the training data and the ground truth data. After training, the systems and methods may use the neural network to detect occluded objects within an environment. For instance, while a vehicle is navigating, the vehicle may process sensor data using the neural network. The neural network may then output data indicating whether an object is located within the environment and occluded from view of the vehicle. In some examples, the neural network may further output additional information associated with the occluded object.