We live in a vast ocean of scenes. Some of these are experienced first-hand as part of our ongoing behavior, others are obtained second-hand from one of the reservoirs of publically available image depositories at our fingertips. Regardless of their source, scenes are ubiquitous, and scene understanding refers to the extraction of content from these uniquely complex images. Behavioral scientists view scenes as the ultimate challenge in a steady march towards using and understanding more complex and realistic visual stimuli. Computer scientists look to scene understanding in order to develop more powerful image tagging and retrieval capabilities, so as to make more accessible this ocean of scene information. The papers in this collection bring together perspectives from behavioral and computer science, highlighting the methods and techniques that each bring to the understanding of scene understanding.
Further complicating the study of scene understanding is the multi-faceted nature of the research topic. Scene understanding means different things to different researchers. For some, it is scene categorization—the ability to rapidly recognize a scene as a street, a forest, or a room. For others it is scene segmentation—using knowledge of the scene category to recognize all of its constituent regions and objects. Still others equate scene understanding with scene perception—determining the relationships between the objects in a scene, between the objects and surfaces, and what elements belong in the perceptual foreground and background. And finally there is the scene’s gist, knowing what the scene is about—its “story”. This might be as simple as the category of the scene (a beach), or a far more elaborated description (a boy and a girl preventing a dog from jumping on the sand castle that they just built). Do these different facets of scene understanding reflect different independent dimensions, or do they represent different stages in the understanding of a scene? In this Research Topic we will explore what it means to understand a scene, and how the different methods and techniques developed by different disciplines might be combined and used to accelerate the understanding of scene understanding.
We live in a vast ocean of scenes. Some of these are experienced first-hand as part of our ongoing behavior, others are obtained second-hand from one of the reservoirs of publically available image depositories at our fingertips. Regardless of their source, scenes are ubiquitous, and scene understanding refers to the extraction of content from these uniquely complex images. Behavioral scientists view scenes as the ultimate challenge in a steady march towards using and understanding more complex and realistic visual stimuli. Computer scientists look to scene understanding in order to develop more powerful image tagging and retrieval capabilities, so as to make more accessible this ocean of scene information. The papers in this collection bring together perspectives from behavioral and computer science, highlighting the methods and techniques that each bring to the understanding of scene understanding.
Further complicating the study of scene understanding is the multi-faceted nature of the research topic. Scene understanding means different things to different researchers. For some, it is scene categorization—the ability to rapidly recognize a scene as a street, a forest, or a room. For others it is scene segmentation—using knowledge of the scene category to recognize all of its constituent regions and objects. Still others equate scene understanding with scene perception—determining the relationships between the objects in a scene, between the objects and surfaces, and what elements belong in the perceptual foreground and background. And finally there is the scene’s gist, knowing what the scene is about—its “story”. This might be as simple as the category of the scene (a beach), or a far more elaborated description (a boy and a girl preventing a dog from jumping on the sand castle that they just built). Do these different facets of scene understanding reflect different independent dimensions, or do they represent different stages in the understanding of a scene? In this Research Topic we will explore what it means to understand a scene, and how the different methods and techniques developed by different disciplines might be combined and used to accelerate the understanding of scene understanding.