<![CDATA[Research related to facial expression recognition (FER) has become a significant topic of interest in the field of computer vision due to its broad applications. Artificial intelligence technologies, such as deep learning, have been applied in FER research. The use of deep learning models in FER requires a dataset for training, which plays a crucial role in determining the performance of deep learning. However, the available FER datasets often require preprocessing before being processed using deep learning. In this study, a comparison of contrast adjustment preprocessing methods was conducted using Histogram Equalization (HE) and Contrast Limited Adaptive Histogram Equalization (CLAHE). Subsequently, the dataset images were used with a fine-tuned deep learning model, specifically AlexNet, to classify them according to the categories of human facial expressions. The objective of this research is to determine the superior contrast adjustment method for FER dataset images in improving the performance of the deep learning model employed. The CK+ dataset (The Extended Cohn-Kanade) and KDEF dataset (The Karolinska Directed Emotional Faces) were used in this study. The results indicate that the CLAHE method outperforms HE in both the CK+ and KDEF datasets. In the CK+ dataset, the CLAHE method achieved an average accuracy of 93.21%, while the average accuracy of the HE method was 91.50%. For the KDEF dataset, the average accuracy of the CLAHE method was 88.35%, compared to 84.70% for the HE method.]]>
