What do We Learn by Semantic Scene Understanding for Remote Sensing imagery in CNN framework?
This work addresses remote sensing image analysis, an incremental improvement for domain-specific applications in geospatial and environmental monitoring.
The study investigated whether deep convolutional neural networks (DCNN) designed for natural images are effective for remote sensing scene recognition, finding that complex scenes require deeper networks and multi-scale perception, with multi-objective joint semantic support being crucial.
Recently, deep convolutional neural network (DCNN) achieved increasingly remarkable success and rapidly developed in the field of natural image recognition. Compared with the natural image, the scale of remote sensing image is larger and the scene and the object it represents are more macroscopic. This study inquires whether remote sensing scene and natural scene recognitions differ and raises the following questions: What are the key factors in remote sensing scene recognition? Is the DCNN recognition mechanism centered on object recognition still applicable to the scenarios of remote sensing scene understanding? We performed several experiments to explore the influence of the DCNN structure and the scale of remote sensing scene understanding from the perspective of scene complexity. Our experiment shows that understanding a complex scene depends on an in-depth network and multiple-scale perception. Using a visualization method, we qualitatively and quantitatively analyze the recognition mechanism in a complex remote sensing scene and demonstrate the importance of multi-objective joint semantic support.