匿名痤疮面部数据集
收藏资源简介:
本研究创建的匿名痤疮面部数据集包含1473张不同严重程度的痤疮面部图像,旨在解决生物医学领域中病理人脸图像数据集的不足问题。数据集通过StyleGAN2算法生成,能够模拟轻度、中度和重度痤疮面部图像,图像分辨率高达1024x1024。创建过程中,首先收集了来自ACNE04数据集和谷歌图像的痤疮面部图像,然后通过深度学习技术进行预处理和增强,确保图像质量。该数据集不仅支持深度学习模型的训练和验证,还可用于教育和研究,帮助解决痤疮诊断和治疗中的实际问题。
The anonymized facial acne dataset created in this study contains 1,473 facial acne images with varying degrees of severity, aiming to address the shortage of pathological facial image datasets in the biomedical field. This dataset is generated using the StyleGAN2 algorithm, which can simulate facial acne images of mild, moderate and severe degrees, with an image resolution of up to 1024×1024. During its creation, facial acne images were first collected from the ACNE04 dataset and Google Images, followed by preprocessing and augmentation via deep learning techniques to ensure image quality. This dataset not only supports the training and validation of deep learning models, but can also be used for education and research, helping to solve practical problems in acne diagnosis and treatment.

- 1Generative Adversarial Networks for anonymous Acneic face dataset generationLISSI实验室 · 2022年



