| Technical Name | Study the Use of Image Generation Techniques to Improve the Performance of AI Assisted Diabetic Retinopathy Diagnoses | ||
|---|---|---|---|
| Project Operator | National Yang Ming University | ||
| Project Host | 賴穎暉 | ||
| Summary | A large number of images of diabetic retinopathy are essential for AI recognition systems, but this often causes clinical personnel to incur high costs in collecting these images. This technology uses the GAN approach to synthesis the training images for AI recognition system, and try to reduce the time and system development costs for clinical personnel in collecting these images and label data. |
||
| Scientific Breakthrough | Our technology can increase the training dataset by the GAN technology; meanwhile, these synthesized images can further improve the performance of AI-based image recognition system. |
||
| Industrial Applicability | Our technology can reduce the time and cost of collecting and organizing data for developing artificial intelligence model through data augmentation. This technology can be applied to save development costs for an artificial intelligence system that needs to be trained in big data. |
||
| Keyword | Artificial Intelligence Machine Learning Deep Learning Convolutional Neural Network Generative Adversarial Network Computer Aided Diagnosis Data Augmentation Diabetic Retinopathy Image Recognition Computer Vision | ||
| Notes | |||
- Contact
- lin3000y@hotmail.com
other people also saw