Interictal high-frequency oscillations (HFOs) hold promise as potential biomarkers for identifying seizure onset zone (SOZ) and monitoring disease activity in patients with drug-resistant epilepsy. However, manual detection of HFOs as well as SOZ in the brain is time-consuming, and the lack of quantitative diagnostic criteria hinders their clinical utilization. To address these challenges, we have developed an automated threshold-based HFO detection and SOZ localization system from electrocorticogram (ECoG) data and investigated the relevance of channel-wise HFO population information in the excised region of four patients who exhibited good postoperative seizure control. Our HFO detection model can accurately predict HFOs from ECoG data and extract channel-wise HFO population information for a new patient without any manually annotated sample. In addition, a general threshold-based solution is provided to automatically localize SOZs from the channel-wise population of detected HFOs. Results show that our SOZ detector performs well in localizing SOZ with accuracy of 87.09%. Our proposed method has the potential to enhance clinical research by supporting accurate HFO detection and precise localization of the epileptogenic zone, contributing to successful epilepsy surgeries.