Boundary detection, a crucial pre-processing step in remote sensing applications such as change detection, monitoring, urbanization, and target detection, aims to identify contours separating objects in images. To address the challenges posed by evolving spatial, spectral, and temporal resolutions of remote sensing data, incorporating multi-features like shape, color, and texture has been explored. However, this approach requires extensive training and engineering of limitless features to cope with the constant development of new imagery.This seminar introduces a supervised boundary detection scheme that eliminates the need for engineered features and manual parameter settings. The proposed framework aims to integrate complementary information from multiple spectral bands of remotely sensed images to create a composite image. This composite image enhances perception, reinforces interpretation, and allows self-learning and customization for detection and fusion tasks.