Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This study conducts a comprehensive analysis of synthetic images generated by different methods compared to real images, focusing on high-dimensional feature space, low-level color statistics, and the model training process. By investigating how these synthetic images can be effectively mixed with real data, the research provides a framework for evaluating the quality of synthetic data before its use in training. The key finding reveals that careful assessment and integration strategies can significantly enhance the reliability and safety of image classification models that leverage synthetic data.
Synthetic images can be safely integrated into training processes, but only if their quality is rigorously assessed against real data.
Recently, the societal implementation of high-performance image classification models has expanded rapidly. While these models require vast amounts of training data to improve performance, securing sufficient real images is often impractical. As a means to compensate for this shortage, the use of synthetic data is becoming widespread. However, synthetic images are not necessarily equivalent to real images for training purposes. This study systematically analyzes the differences between two types of synthetic images created by different generation methods and real images from three perspectives: high-dimensional feature space, low-level statistics in color space, and the model training process. Furthermore, it experimentally verifies how synthetic data should be utilized by considering realistic data mixing scenarios. This enables the proposal of an evaluation and application strategy for performing preliminary assessments on synthetic images of unknown quality and safely incorporating them into training. This research aims to contribute to enhancing the reliability and safety of image classification models utilizing synthetic images.