Performance comparison of various deep learning models across precision, recall, F1 score and AUC metrics. The table highlights the effectiveness of different architectures, including CNN, LSTM, their combination (CNN + LSTM), Vision Transformer and the proposed HIMP model. The HIMP model outperforms all others, achieving the highest scores in all evaluation metrics, particularly with an F1 score of 91.20% and an AUC of 95.36%