This report presents an approach to the Capsule Vision 2024 Challenge, which focuses on multi-class abnormality classification in video capsule endoscopy (VCE) frames. The challenge aims to develop AI models capable of automatically classifying abnormalities captured in VCE video frames into ten distinct categories. To address this, a deep learning-based solution was developed. The model was trained on a diverse dataset comprising 37,607 VCE frames and validated on an additional 16,132 frames. Through extensive hyperparameter tuning, data augmentation, and sampling techniques, the model achieved a mean AUC of 0.98 and a balanced accuracy of 0.83 on the validation dataset. These results demonstrate the effectiveness of this approach in accurately classifying VCE abnormalities, potentially reducing the diagnostic burden on clinicians and improving patient outcomes.