In order to meet the requirement of fine-grained viewpoint prediction for design drawings, a deep learning-based viewpoint prediction model is developed to achieve heat map generation, feature element recognition and interactive and instantaneous detection of design solutions, and the attentional influencing factors are discussed. A saliency map is introduced to model the visual attention allocation mechanism, and a fully convolutional networks-based Image Viewpoint Prediction Model (IVPM) is proposed to overcome the limitations of eye tracker tests. The model has excellent temporal performance after training on the Graphics Design Importance (GDI) dataset, and experiments validate that low-level attributes of images are the main influences on design attention. The IVPM can be applied to image creation, poster design, packaging design, product design and interface design, and is a useful reference for the creative design field.
Student, Bachelor of Arts, Winchester School of Art, University of Southampton, Hampshire, United Kingdom
Attention Management, Eye Tracking, Viewpoint Prediction, Design, Deep Learning
This presenter hasn’t added media.
Request media and follow this presentation.