DSTC8 Track Proposal: Dialog State Tracking for Conversational Image Editing

We propose a new challenge for the dialog research community: dialog state tracking for conversational image editing. Conversational image editing [1] is a novel application domain that combines language and vision as well as action execution in task-oriented scenarios. The dialog application goes beyond traditional slot-filling systems for restaurant search or movie booking. In addition, it is also more challenging compared to visual dialog [2], since in conversational image editing the image content is being changed as the conversation unfolds.