3. Once the dataset annotations were finalized, we have utilized the best-performing model from each learning approach presented in [9], [53], [54] to benchmark our proposed dataset and to compare the results with SOTA approaches. All experiments were done with a 5-fold cross-validation ...