Image Identification Model Training Using Intermediary Clear Reference Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image identification systems using computational images from light-field cameras face challenges in visual recognition due to blurring, making it difficult to assign accurate correct answer labels and resulting in deteriorated learning efficiency, while also requiring effective privacy protection in environments like homes or indoors.
Innovation Solution
A method involving a learning device that acquires computational imaging information from a first camera with blurring, generates a third image without blurring or with smaller blurring, and uses machine learning to create an image identification model, ensuring accurate label assignment and improved learning efficiency while protecting privacy by utilizing cameras like multi-pinhole or light-field cameras.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If computational imaging information is used to protect privacy, then privacy protection is improved, but learning efficiency deteriorates due to difficulty in assigning accurate correct answer labels
Solution Approach 1:
A third camera that captures clear images is introduced as an intermediary to provide accurate reference images for label assignment. The clear images from the third camera serve as a mediator between the blurred computational images and the required accurate labels, enabling efficient supervised learning while maintaining privacy protection through the blurred images.
2Reliability
If images with blurring are used for privacy protection, then privacy protection is improved, but image identification accuracy deteriorates
Solution Approach 1:
The system segments the imaging function into multiple cameras with different characteristics: the first camera captures blurred computational images for privacy protection, while the third camera captures clear images for accurate identification. This segmentation allows each camera to fulfill its specific function without compromise.
Solution Approach 2:
The third camera performs preliminary action by capturing clear reference images in advance that are used to assign accurate correct answer labels for training the identification model. This preliminary capture of clear images enables subsequent accurate identification while the first camera continues to capture privacy-protecting blurred images.
Data Source
AI summary
Provided is a learning device that acquires computational imaging information of a computational imaging camera that captures an image with blurring; acquires a normal image captured by a normal camera that captures an image without blurring or an image with blurring smaller than that of the computational imaging camera, and a correct answer label assigned to the normal image; generates an image with blurring based on the computational imaging information and the normal image; and performs machine learning using the image with blurring and the correct answer label to create an image identification model for identifying an image captured by the computational imaging camera.