Synthetic Fixation Maps for Low-Cost Pedestrian Detection Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated driving systems face challenges in quickly and accurately detecting pedestrians due to the difficulty in training and testing deep learning algorithms, which require extensive and costly sensor data and annotations, and struggle to match human perception in understanding scene gist and object localization.
Innovation Solution
The development of synthetic saliency maps, generated by creating intermediate images with random points and applying Gaussian blur, allows for the creation of low-resolution labels that can be used to train and test deep neural networks, reducing computational power and time required for object detection, and mimicking human perception without exhaustive data collection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning algorithms are trained using extensive sensor data and annotations to improve pedestrian detection accuracy, then object detection accuracy is improved, but training time and cost increase significantly
Solution Approach 1:
The patent creates synthetic saliency maps that copy and simulate the essential characteristics of human visual fixation patterns without requiring actual human annotation. By generating artificial saliency maps through computational algorithms that mimic human attention distribution, the system obtains training labels that replicate genuine human perception data, thereby reducing the time and cost of data collection while maintaining detection accuracy
Solution Approach 2:
The system performs self-annotation by automatically generating its own training labels through computational algorithms. Instead of requiring external human annotators, the system uses automated saliency map generation methods to create training data, enabling self-service data preparation that eliminates the time-consuming manual annotation process while preserving the essential patterns needed for accurate pedestrian detection
2Measurement precision
If deep learning algorithms are trained using extensive sensor data and annotations to improve pedestrian detection accuracy, then object detection accuracy is improved, but training cost increases significantly
Solution Approach 1:
The patent creates synthetic saliency maps that copy and simulate the essential characteristics of human visual fixation patterns without requiring actual human annotation. By generating artificial saliency maps through computational algorithms that mimic human attention distribution, the system obtains training labels that replicate genuine human perception data, thereby reducing the time and cost of data collection while maintaining detection accuracy
Solution Approach 2:
The system performs self-annotation by automatically generating its own training labels through computational algorithms. Instead of requiring external human annotators, the system uses automated saliency map generation methods to create training data, enabling self-service data preparation that eliminates the time-consuming manual annotation process while preserving the essential patterns needed for accurate pedestrian detection
3Measurement precision
If automated driving systems use high-resolution sensor data for object detection, then detection accuracy is improved, but computational power and processing time increase
Solution Approach 1:
The patent extracts and emphasizes only the most critical visual information by generating saliency maps that highlight regions of interest in the sensor data. Instead of processing the entire high-resolution image, the system extracts salient regions that contain the most important information for pedestrian detection, thereby reducing computational load while maintaining detection accuracy
Solution Approach 2:
The patent segments the visual field into salient and non-salient regions through saliency map generation. By dividing the image processing task into focused regions of interest rather than processing the entire high-resolution image uniformly, the system reduces computational power requirements while preserving detection accuracy in the most critical areas
4Measurement precision
If automated driving systems aim to match human perception capabilities, then pedestrian detection accuracy is improved, but data collection and annotation complexity increase
Solution Approach 1:
The system performs self-annotation by automatically generating its own training labels through computational algorithms. Instead of requiring external human annotators, the system uses automated saliency map generation methods to create training data, enabling self-service data preparation that eliminates the time-consuming manual annotation process while preserving the essential patterns needed for accurate pedestrian detection
Solution Approach 2:
The patent creates synthetic saliency maps that copy and simulate the essential characteristics of human visual fixation patterns without requiring actual human annotation. By generating artificial saliency maps through computational algorithms that mimic human attention distribution, the system obtains training labels that replicate genuine human perception data, thereby reducing the time and cost of data collection while maintaining detection accuracy
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
This approach significantly reduces the time and cost of training and testing deep learning algorithms, improves object detection accuracy, and enhances the ability to predict object locations and scales, enabling more efficient and effective automated pedestrian detection in autonomous vehicles.
Implementation Method 1
applying a Gaussian blur to the intermediate image to produce a blurred intermediate image
Data Source
AI summary
The disclosure extends to methods, systems, and apparatuses for automated fixation generation and more particularly relates to generation of synthetic saliency maps. A method for generating saliency information includes receiving a first image and an indication of one or more sub-regions within the first image corresponding to one or more objects of interest. The method includes generating and storing a label image by creating an intermediate image having one or more random points. The random points have a first color in regions corresponding to the sub-regions and a remainder of the intermediate image having a second color. Generating and storing the label image further includes applying a Gaussian blur to the intermediate image.