Multiclass semantic segmentation outcomes are illustrated in Desk 4. As a result of significant amount of lessons and class imbalance, the performance of multiclass semantic segmentation is less satisfactory. Pre-properly trained DeepLabV3 has proven the best general performance between all versions. In addition, weighted F1 scores also are minimal as https://self-discovery44332.diowebhost.com/88047274/the-fact-about-aerial-survey-bangladesh-that-no-one-is-suggesting