Photographs convey the stories of photographers to the audience. However, this story-telling aspect of photography is easily distracted by visual clutter. Informed by a pilot study, we identified the kinds of clutter that amateurs frequently include in their photos. We were thus inspired to develop DeclutterCam, a photographic assistant system that incorporates novel user interactions and AI algorithms for photographic decluttering. Clutter elements are detected by an aesthetic quality evaluation algorithm and are highlighted so that users can interactively identify distracting elements. A GAN-based iterative clutter removal tool enables users to test their photographic ideas in real-time. User studies with 32 photography beginners demonstrate that our system provides flexible interfaces, accurate algorithms, and immediate feedback that allow users to avoid clutter and explore more photographic ideas. Evaluations by photography experts show that users can take higher-quality photos that better convey the intended story using our system.
翻译:照片将摄影师的故事传达给观众。 然而,这种讲述摄影故事的方面很容易被视觉乱七八糟的视觉乱七八糟地分散开来。 通过试点研究,我们发现了业余人士经常在其照片中包含的杂乱无章的画面。因此,我们被启发开发了Declutter Cam,这是一个包含新颖用户互动和照片淡化的AI算法的摄影助理系统。通过审美质量评估算法检测到各种刻板元素,并被突出显示用户能够交互识别分散的元素。一个基于GAN的迭接式清除工具使用户能够实时测试他们的照片想法。与32名摄影创始人进行的用户研究显示,我们的系统提供了灵活的界面、准确的算法和即时反馈,使用户能够避免杂乱和探索更多的摄影想法。摄影专家的评估表明,用户可以采取更高质量的照片,用我们的系统更好地传达预期的故事。