Model Training System Automatic Background Replacement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current deep learning platforms face inefficiencies when manually processing and uploading synthesized images for model training, as users must perform offline background removal and synthesis, which is time-consuming and inefficient.
Innovation Solution
A model training method and system that automatically detects on-image marks to perform background replacement, generating new images with different backgrounds and creating training data for improved model training efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If manual offline synthesis and upload is used to expand training data, then training data diversity is improved, but training efficiency deteriorates
Solution Approach 1:
The system enables self-service by allowing the model training platform to automatically perform background removal and image synthesis without requiring external image processing software or manual user operations. The platform integrates these functions natively, enabling it to serve itself in generating diverse training data efficiently
Solution Approach 2:
The patent merges the background removal function and image synthesis function into a single integrated process within the model training platform. By combining these previously separate operations (manual background removal + manual synthesis + manual upload) into one automated workflow, the system achieves both data diversity and training efficiency
2Ease of manufacture
If additional image processing software is used for background removal and synthesis, then image processing capability is improved, but system complexity increases
Solution Approach 1:
The model training platform is designed with multi-functionality, serving both as a model training environment and an integrated image processing system. By incorporating background removal and synthesis capabilities directly into the platform, it eliminates the need for separate specialized software tools, thereby reducing system complexity while maintaining comprehensive image processing capability
3Manufacturing precision
If manual offline synthesis is performed, then image synthesis quality is improved, but processing time increases
Solution Approach 1:
The system performs preliminary action by pre-integrating the background removal and synthesis algorithms into the model training platform before actual training data generation is needed. This preparation enables rapid, automated processing of training images without requiring manual intervention during the actual data generation phase, thus maintaining quality while reducing processing time
Data Source
AI summary
A model training method and a model training system are disclosed. The method includes the following. A first image with an on-image mark is obtained. In response to the on-image mark of the first image, an automatic background replacement is performed on the first image to generate a second image. A background image of the second image is different from a background image of the first image. Training data is generated according to the second image. An image identification model is trained by using the training data.