Annotation Correction for Antenna Detection in Aerial Imagery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing technologies face challenges in processing load and accuracy when detecting and correcting annotations for predetermined objects in images, particularly in distinguishing antenna devices from outdoor backgrounds.
Innovation Solution
An information processing apparatus that acquires images with annotations, specifies regions based on annotation criteria, performs edge detection within those regions, and corrects annotations to align with detected edges, using machine learning to enhance object detection and angle calculation for antenna devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If edge detection is performed on the entire image to correct annotations, then annotation accuracy is improved, but processing load increases
Solution Approach 1:
The image is divided into multiple regions based on annotation density criteria. The edge detection unit performs edge detection preferentially in specified regions where annotations meet predetermined criteria (e.g., regions with high annotation density or specific spatial patterns). This segmentation approach allows accurate annotation correction in critical areas while avoiding unnecessary processing in other regions, thereby reducing overall processing load while maintaining annotation accuracy where it matters most.
2Measurement precision
If machine learning models are trained with more annotation data, then detection accuracy for antenna devices is improved, but data preparation time increases
Solution Approach 1:
The system performs preliminary annotation correction using the edge detection unit before machine learning model training. By automatically correcting annotations to align with detected edges in advance, the system prepares higher-quality training data more efficiently. This preliminary action reduces the time needed for manual annotation verification and ensures that the machine learning model trains on accurate data, improving detection accuracy while reducing overall data preparation time.
3Manufacturing precision
If manual annotation correction is performed to align annotations with object edges, then annotation precision is improved, but operation time increases
Solution Approach 1:
The system replaces manual mechanical annotation correction with an automated computational process. The edge detection unit automatically detects edges in the image, and the annotation correction unit programmatically adjusts annotation positions and orientations to align with detected edges. This substitution of manual operation with automated image processing algorithms achieves high annotation precision while dramatically reducing the time required compared to manual correction methods.
Data Source
AI summary
[Problem] Provided is a novel information processing technology relating to a predetermined object in an image.[Solving Means] An information processing apparatus includes: an image acquisition unit that acquires an image used as teacher data for machine learning, the image being with one or a plurality of annotations for showing a position in the image at which a predetermined object is shown; a region specification unit that specifies a region in which the one or the plurality of annotations satisfy a predetermined criterion in the image; an edge detection unit that preferentially detects edges in the specified region or a range set on a basis of the region; and an annotation correction unit that corrects the annotations so as to be along the detected edges.