We developed an endoscopic image artifact classifier for blur and saturation using generated images to enable real-time quality assessment during endoscopic examinations. After training StyleGAN-ADA with real images free of artifacts, we fine-tuned th...
We developed an endoscopic image artifact classifier for blur and saturation using generated images to enable real-time quality assessment during endoscopic examinations. After training StyleGAN-ADA with real images free of artifacts, we fine-tuned the model using real images with single or dual artifacts. As a result, the classifier trained with single-artifact generated images demonstrated the highest performance in artifact classification.