Most Arabic interactions in media (TV, radio, etc), and on the internet (social media, forums) are produced in local dialects. Dialectal Arabic (DA) is significantly different from the formal Arabic language, known as Modern Standard Arabic (MSA). Especially on social media, we observe various dialects and free writing forms that make the Natural Language Processing task more complicated. Misogyny, which is defined as the hate towards women, or the notion that men are far superior to women, has spread across a range of social media platforms, becoming a global epidemic. Women in the Arab world face a wide range of online misogyny, which sadly reinforces and excuses gender inequality, violence against women, and women’s undervaluation. From here came the challenge of misogyny detection in Arabic dialects. Therefore, the Arabic Misogyny Identification (ArMI) task is the first shared task that attempts to address the issue of automatic identification of Arabic online misogyny. The ArMI shared task attempts to identify the misogynistic content and recognize the different misogynistic behaviors in a collection of Arabic (MSA/dialectal) tweets. In this paper, we describe the system submitted for the ArMI shared task on misogyny detection in Arabic dialects. In this challenge, we conducted experiments on MarBERT, a BERT-based model that focused on both Dialectal Arabic (DA) and MSA. We utilized the MarBERT pretrained model that we fine-tuned on the provided training set after applying our preprocessing strategies.. The rest of the paper is organized as follows: in Section 2 we introduce ArMI dataset. In Section 3, we describe our approach in tackling the problem. Then, in Section 4 we provide and discuss the results of the proposed method on Subtask-1. Section 5 concludes our work throughout this shared task.