Aerial Image Object Detection via Segmentation and Training Augmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting objects from aerial imagery face challenges in accuracy and efficiency, particularly when dealing with large areas and high-resolution images, which increase computing complexity and make manual identification infeasible.
Innovation Solution
A system and method that involves obtaining aerial images, classifying regional images into target and non-target classes using a trained classifier, and recognizing target objects by adjusting parameters such as brightness, contrast, color saturation, resolution, and rotation angle of training images, allowing for efficient detection of objects like oil palm trees or other features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the resolution of aerial images is increased to improve detection accuracy, then measurement precision is improved, but device complexity and computing complexity increase
Solution Approach 1:
The system divides the large-scale aerial image into multiple smaller regional images for processing. This segmentation approach reduces the computational complexity of processing high-resolution images while maintaining detection accuracy by applying the trained classifier to smaller, more manageable regions that can be handled more efficiently.
2Measurement precision
If manual identification is used to improve detection accuracy, then measurement precision is improved, but productivity decreases
Solution Approach 1:
The system replaces manual identification with an automated classifier trained on training images. This substitution of mechanical/manual processes with an automated computational system maintains high detection accuracy while dramatically improving productivity by enabling rapid processing of large-scale aerial imagery without human intervention.
3Productivity
If conventional object detection methods are used to improve processing speed, then productivity is improved, but measurement precision deteriorates
Solution Approach 1:
The system performs preliminary training of the classifier using training images before actual detection. This preliminary action prepares the system to quickly and accurately process aerial images by pre-learning the characteristics of target objects, thereby achieving both high processing speed and high detection accuracy simultaneously.
Data Source
AI summary
Methods and systems for detecting objects from aerial imagery are disclosed. The method includes obtaining an image of an area, obtaining a plurality of regional aerial images from the image of the area, classifying the plurality of regional aerial images as a first class or a second class by a classifier, wherein: the first class indicates a regional aerial image contains a target object, the second class indicates a regional aerial image does not contain a target object, and the classifier is trained by first and second training data, wherein the first training data include first training images containing target objects, and the second training data include second training images containing target objects obtained by adjusting at least one of brightness, contrast, color saturation, resolution, or a rotation angle of the first training images; and recognizing a target object in a regional aerial image in the first class.