SAM Based Automatic Object Segmentation Based on User Selection

Visual segmentation is an essential approach for discriminating and detecting objects in complex situations in computer vision. However, existing methods are often limited to segmenting the entire image or a selected area. In this work, there is no need for unique model building. Segmentation is performed over the similar objects of the object to be segmented. This paper presents a new method automatically segments all areas identical to the selected objects. Our proposed method segments all similar objects by using the area features of the selected objects. Existing segmentation techniques usually focus only on specific regions or the entire image. For this purpose, the proposed approach is based on the SAM algorithm and is realisable with a user-friendly interface. Even in complex cases, the proposed approach enables high-accuracy segmentation of similar elements. The results of the study show how practical and valuable our approach is.