Superpixel-Based Multiscale CNN Approach Toward Multiclass Object Segmentation From UAV-Captured Aerial Images

Unmanned aerial vehicles (UAVs) are promising remote sensors capable of reforming sensing applications. However, for artificial intelligence (AI)-guided tasks, like land cover mapping and ground-object mapping, most deep learning-based architectures fail to extract scale-invariant features, resulting in poor performance accuracy. In this context, the article proposes a superpixel-aided multiscale convolutional neural network (CNN) architecture avoid misclassification complex urban images.The proposed framework is two-tier segmentation architecture. first stage, superpixel-based simple linear iterative cluster (SLIC) algorithm produces superpixel images with crucial contextual information. The second stage comprises CNN that uses these information-rich features predicting object class each pixel. Two UAV-image-based image datasets: NITRDrone dataset xmlns:xlink="http://www.w3.org/1999/xlink">urban drone dataset (UDD) considered perform experimentation. model outperforms state-of-the-art methods an intersection union (IoU) 76.39% 86.85% on UDD NITRDrone datasets, respectively. Experimentally obtained results prove performs superior by achieving better accuracy challenging scenarios.