Automatic Annotation Transfer via Keypoint Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems for UAVs face challenges in automatically transferring annotations from low-resolution images to high-resolution images, as the latter are only available after the UAV returns to the ground station, requiring manual reapplication of annotations which is time-consuming and prone to user errors.
Innovation Solution
The system uses processing circuitry to match keypoints in image patches between low-resolution and high-resolution images, allowing for automatic transfer of annotations based on corresponding image patches, even with differences in zoom, angle, or orientation, thereby synchronizing annotations without user intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If annotations are manually transferred from low-resolution images to high-resolution images, then annotation accuracy is maintained, but time consumption and user effort increase significantly
Solution Approach 1:
The system performs automatic annotation transfer by matching keypoints between low-resolution and high-resolution images, allowing the system to serve itself without requiring manual user intervention for the repetitive task of reapplying annotations to different image resolutions
Solution Approach 2:
The manual mechanical process of copying annotations is replaced with an automated computational system that uses keypoint detection and image patch matching algorithms to automatically transfer annotations, substituting human manual labor with computational processing
2Speed
If annotations are applied to low-resolution images in real-time, then real-time inspection feedback is enabled, but annotation synchronization to high-resolution images is delayed
Solution Approach 1:
The system performs preliminary keypoint detection and establishes matching relationships between low-resolution and high-resolution images in advance, so that when annotations are created on low-resolution images, their transfer to high-resolution images can occur automatically and immediately without waiting for manual intervention
Solution Approach 2:
The system creates a feedback loop where annotations made on low-resolution images are automatically transferred to high-resolution images, and the synchronization status can be monitored, ensuring that annotation information is consistently maintained across both image resolutions
3Measurement precision
If manual annotation transfer is performed, then annotation precision is maintained, but user errors increase due to repetitive manual work
Solution Approach 1:
The system eliminates the need for manual annotation transfer by implementing self-service automatic transfer mechanisms that use keypoint matching and image patch comparison to accurately relocate annotations from low-resolution to high-resolution images without human intervention
Solution Approach 2:
The system creates accurate copies of annotations by detecting corresponding keypoints between images and reproducing annotation positions and characteristics in the target high-resolution image, ensuring annotation precision is maintained through systematic copying rather than manual recreation
Data Source
AI summary
In some examples, a computing device includes a display, processing circuitry configured to present a first image via the display, and an input device configured to receive user inputs. The processing circuitry is further configured to determine an annotation to the first image based on the user inputs and determine an image patch in the first image overlapping with the annotation. The processing circuitry is also configured to determine, in the image patch, a first set of keypoints associated with the annotation and match the first set of keypoints in the first image to a second set of keypoints in the corresponding image patch in the second image. The processing circuitry is configured to determine a position of the corresponding image patch in the second image based on matching the first set of keypoints to the second set of keypoints and apply the annotation to the second image.