Defect detection system

The computing system generates a training dataset for training the prediction model to detect defects present in the target surface of the target sample, and trains the prediction model based on the training dataset to detect defects present in the target surface of the target sample. A computing system generates a training dataset by identifying a set of images for training a prediction model, the set of images including a first subset of images. The deep learning network generates a second subset of images for subsequent marking based on the image set comprising the first subset of images. The deep learning network generates a third subset of images for tagging based on an image set including the first subset of images and the tagged second subset of images. The computing system continues the process until a threshold number of tagged images are generated. 计算系统生成用于训练预测模型以检测目标样本的目标表面中存在的缺陷的训练数据集，以及基于训练数据集训练预测模型以检测目标样本的目标表面中存在的缺陷。计算系统通过识别用于训练预测模型的图像集来生成训练数据集，图像集包括第一图像子集。深度学习网络基于包括第一图像子集的图像集，生成用于后续标记的第二图像子集。深度