Object Detection via Geometric Image Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Deep learning models for object detection often fail when the target object has high variability in posture, leading to detection failures and the need for repeated training.
Innovation Solution
The method involves performing geometric transformations on an initial image to generate multiple secondary images, which are combined with the original image to create a richer combination image, enhancing the detection accuracy by inputting this combination into a trained deep learning model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If geometric transformation operations are performed on the first image to obtain at least one second image, then detection accuracy is improved, but processing time and computational complexity increase
Solution Approach 1:
The patent applies geometric transformation operations (such as rotation, flipping, scaling) to the input image before object detection to generate multiple transformed images. These transformations are performed in advance to create a more comprehensive set of image variations that help the deep learning model detect objects with different postures and orientations, thereby improving detection accuracy without requiring repeated model training.
Solution Approach 2:
The patent extends the detection process from a single image to multiple dimensions by creating transformed images through geometric operations. This adds temporal and spatial dimensions to the detection process, allowing the model to analyze the same scene from multiple transformed perspectives, which significantly improves detection accuracy for objects with varying postures.
2Measurement precision
If geometric transformation operations are performed on the first image to obtain at least one second image, then detection accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies geometric transformation operations (such as rotation, flipping, scaling) to the input image before object detection to generate multiple transformed images. These transformations are performed in advance to create a more comprehensive set of image variations that help the deep learning model detect objects with different postures and orientations, thereby improving detection accuracy without requiring repeated model training.
Solution Approach 2:
The patent extends the detection process from a single image to multiple dimensions by creating transformed images through geometric operations. This adds temporal and spatial dimensions to the detection process, allowing the model to analyze the same scene from multiple transformed perspectives, which significantly improves detection accuracy for objects with varying postures.
3Measurement precision
If the first image and the at least one second image are combined to generate a combination image, then detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent combines the original first image with one or more geometrically transformed second images to create a combination image. This merging process integrates multiple perspectives and variations of the same scene into a single composite image that contains richer information, enabling the deep learning model to detect objects with varying postures and orientations more accurately in a single detection pass.
4Measurement precision
If the first image and the at least one second image are combined to generate a combination image, then detection accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent combines the original first image with one or more geometrically transformed second images to create a combination image. This merging process integrates multiple perspectives and variations of the same scene into a single composite image that contains richer information, enabling the deep learning model to detect objects with varying postures and orientations more accurately in a single detection pass.
Data Source
AI summary
An object detection method, an electronic apparatus and an object detection system are provided. The method is adapted to the electronic apparatus and includes the following steps. A first image is obtained. A geometric transformation operation is performed on the first image to obtain at least one second image. The first image and the at least one second image are combined to generate a combination image. The combination image including the first image and the at least one second image is inputted into a trained deep learning model to detect a target object.