Ovarian carcinoma is the deadliest cancer of the female reproductive system. It is also a heterogeneous disease with five common histotypes: high-grade serous carcinoma (HGSC) accounts for 70% of cases, clear cell ovarian carcinoma (CCOC) for 12%, endometrioid (ENOC) for 11%, low-grade serous (LGSC) for 4%, and mucinous carcinoma (MUC) for 3%. They differ in their cellular morphologies, etiologies, molecular, genetic, and clinical characteristics. Histotype-based treatment is becoming increasingly prevalent with the introduction of PARP inhibitor therapy for patients with HGSC. Ovarian cancer histotype classification by pathologists is associated with challenges in diagnostic reproducibility and interobserver disagreement. Initial diagnosis is performed through histological assessment of hematoxylin & eosin (H&E)-stained sections, but studies have shown that for pathologists without gynecologic pathology-specific training, the interobserver agreement is only moderate. Furthermore, the number of pathologists trained has not kept up with the increasing volume of cancer diagnoses. OCEAN is a scientific competition for developing an artificial intelligence (AI)-based software package for histopathology images of ovarian cancers. Our challenge comprises digitalized samples from 25 centers, with each image falling into one of three categories: normal, an outlier, and one of the five histotypes of ovarian cancer. Participants are asked to develop deep learning methodologies for classifying ovarian cancer histotypes and identifying outliers. Additionally, variations between slide scanners, different tissue processing and staining protocols across various pathology labs, and inter-patient variability can lead to inconsistent color appearances in histopathology sections; therefore, the generalizability of the developed software is a key aspect of the competition that participants need to take into consideration.