Intelligent System for Landscape(ISI) Integrated with Optimized Scene Segmentation Algorithm

Efficient image segmentation with the integration of the scene understanding is essential for the computer vision application, and using GANs to realize the unsupervised training and data augmentation is one of the current research hotspots. In this study, the novel intelligent system for the landscape (ISI) integrated with optimized scene segmentation algorithm is designed and tested. For the designed algorithm, 3 novelties are considered, namely: (1) Novel GANs based image segmentation algorithm is designed, the multimodal image segmentation based on conditional random field deep convolution generation adversarial network (DCGAN) is considered; (2) The novel YOLOv3 based scene understanding model is designed to construct the intelligent system; (3) The landscape image features are combined to make the model more efficient. The experimentation is conducted and the segmentation performance is validated to be efficient.