Camera Blockage Dataset Generation With Chroma Key Boundaries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately detecting camera blockages, such as debris or water on the lens, which can degrade performance and pose safety concerns.
Innovation Solution
The solution involves generating highly accurate synthetic input data representing camera blockages using chroma keying to capture blockage images on a known background, and then superimposing these images onto various real-world backgrounds to create synthetic blocked images with known blockage boundaries. Additionally, a neural network is trained using a combination of synthetic and non-synthetic images to detect blockages and their boundaries accurately.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If real-world blocked camera images are collected for training, then the training data reflects actual blockage scenarios, but it is difficult to obtain sufficient quantities and the blockage boundaries are unknown
Solution Approach 1:
The patent creates synthetic copies of blockage images by combining real blockage photographs with chroma key technology. Real blockage images are captured against green screen backgrounds, then these blockage elements are extracted and superimposed onto various virtual and real camera images to generate synthetic training datasets with known blockage boundaries and abundant quantity.
Solution Approach 2:
The patent introduces chroma key background (green screen) as an intermediary element. This allows real blockage objects to be separated from their original backgrounds, enabling precise boundary identification and flexible recombination with different camera images to create diverse training scenarios while maintaining accurate blockage boundary information.
2Quantity of substance
If synthetic blockage images are generated without chroma keying, then large volumes of training data can be created, but the accuracy of blockage boundaries deteriorates
Solution Approach 1:
The patent performs preliminary chroma key extraction of blockage elements before combining them with camera images. By pre-separating blockage objects from green screen backgrounds and storing them as independent elements, the system ensures that when these elements are combined with various camera images, the blockage boundaries remain precisely defined and easily identifiable for training purposes.
3Reliability
If more real blocked images are collected to improve training data quality, then detection accuracy improves, but the time and resources required for data collection increase
Solution Approach 1:
The patent performs a single preliminary data collection phase where blockage elements are photographed against green screens. These pre-captured blockage elements can then be indefinitely reused and combined with any number of camera images to generate synthetic training datasets, eliminating the need for continuous time-consuming field data collection while maintaining high 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 enables the generation of large volumes of highly accurate input data for training neural networks, improving the detection of camera blockages and ensuring safer and more accurate operation of autonomous vehicles.
Implementation Method 1
performing a chroma keying operation to extract imagery of a blockage from a partial blockage image
Data Source
AI summary
Provided are methods for scalable and realistic camera blockage dataset generation, which can include generating synthetic images depicting a blockage on or near an imaging sensor. The synthetic images may be created by combining one or more chroma key-extracted partial blockage image with one or more background images, the combination of which can provide a scalable blockage dataset. Metadata for each synthetic image can be generated along with the synthetic image, by annotating the portion of the synthetic image represented by the chroma key-extracted partial blockage image as constituting blockage. The synthetic images can be used to increase the accuracy of machine learning models trained to identify blockage by increasing the volume of data available for such training.