Image Segmentation for Synthetic Training Data Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In new scenarios or fields, there is often a lack of open-source training sample data for machine learning models, leading to high manual costs, low efficiency, and poor quality of collected data.
Innovation Solution
An image generation method that segments a target depth image to obtain a target object template image, which is then superimposed onto various background images to create a large quantity of training images with different backgrounds, thereby reducing costs and improving data quality and training efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If manual data collection is used to obtain training sample data for new scenarios, then data can be collected, but manual costs are high, efficiency is low, and data quality is poor
Solution Approach 1:
The patent uses depth image segmentation to create accurate copies of target objects from a single source image, then synthesizes multiple training samples by combining these object copies with different background images. This copying approach eliminates manual data collection while maintaining high data quality and enabling efficient automated generation of large-scale training datasets.
2Quantity of substance
If manual data collection is used to obtain training sample data, then data can be collected, but manual costs are high
Solution Approach 1:
The patent uses depth image segmentation to create accurate copies of target objects from a single source image, then synthesizes multiple training samples by combining these object copies with different background images. This copying approach eliminates manual data collection while maintaining high data quality and enabling efficient automated generation of large-scale training datasets.
3Quantity of substance
If a small amount of valid sample data is used for training, then training can be performed, but training efficiency and accuracy of the image processing model are limited
Solution Approach 1:
The patent uses depth image segmentation to create accurate copies of target objects from a single source image, then synthesizes multiple training samples by combining these object copies with different background images. This copying approach eliminates manual data collection while maintaining high data quality and enabling efficient automated generation of large-scale training datasets.
Solution Approach 2:
The patent introduces background images as an additional dimension to expand the training dataset. By combining segmented object templates with multiple diverse background images, the system transforms a single object image into numerous training samples with varying contexts, significantly increasing data quantity and diversity for model training.
Data Source
AI summary
This application provides an image generation method performed by a computer device. The method includes: obtaining a target depth image including a target object in a real scene, and each pixel point in the target depth image having a depth value; segmenting the target depth image based on the depth value corresponding to each pixel point in the target depth image, to obtain a target object template image including a plurality of pixel points corresponding to the target object in the target depth image; obtaining M background images corresponding to a target scene, the target scene being a scene set associated with an image processing model, and M being an integer greater than or equal to 1; and superimposing the target object template image on the M background images, to generate M target scene images, the target scene images being configured for training the image processing model.