Özet:

One of the most important stages of deep learning-based image classification studies is the data obtaining stage. The data set to train the model must be unique to the task and of adequate quality. Therefore, the process of creating a data set can be a difficult and tired process for researchers. Web mining techniques provide researchers with solutions in creating appropriate data sets that can be used in the studies. Especially in tasks that require a lot of data, such as deep learning, the process of using these techniques can be seriously accelerated. In this context, this study investigates the success of the data set created with the visual grabbing technology for a sample image classification task. In the study, using different CNN models, the sample data set created was trained. The accuracy and other performance standards support that the data set obtained through visual grabbing can be used for image classification tasks.

Anahtar Kelimeler: