Brightness as an Augmentation Technique for Image Classification

Augmentation techniques are crucial for accurately training convolution neural networks (CNNs). Therefore, these have become the preprocessing methods. However, not every augmentation technique can be beneficial, especially those that change image’s underlying structure, such as color techniques. In this study, effect of eight brightness scales was investigated in task classifying a large histopathology dataset. Four state-of-the-art CNNs were used to assess each scale’s performance. The use beneficial all experiments. Among different scales, [0.75–1.00] scale, which closely resembles original images, resulted best geometric yielded better performance than any scale. Moreover, results indicate CNN without applying led considering augmentation. experimental support hypothesis image classification using deep-learning models and do yield gain. Furthermore, significantly degrade model’s when they applied with extreme values. Doi: 10.28991/ESJ-2022-06-04-015 Full Text: PDF