ViP-CNN: Visual Phrase Guided Convolutional Neural Network
This work addresses visual relationship detection, an intermediate task between image captioning and object detection, for researchers in computer vision, though it appears incremental as it builds on existing methods.
The paper tackles visual relationship detection by formulating it as three inter-connected recognition problems and proposes ViP-CNN with a Phrase-guided Message Passing Structure to address them simultaneously, achieving state-of-the-art results in both speed and accuracy.
As the intermediate level task connecting image captioning and object detection, visual relationship detection started to catch researchers' attention because of its descriptive power and clear structure. It detects the objects and captures their pair-wise interactions with a subject-predicate-object triplet, e.g. person-ride-horse. In this paper, each visual relationship is considered as a phrase with three components. We formulate the visual relationship detection as three inter-connected recognition problems and propose a Visual Phrase guided Convolutional Neural Network (ViP-CNN) to address them simultaneously. In ViP-CNN, we present a Phrase-guided Message Passing Structure (PMPS) to establish the connection among relationship components and help the model consider the three problems jointly. Corresponding non-maximum suppression method and model training strategy are also proposed. Experimental results show that our ViP-CNN outperforms the state-of-art method both in speed and accuracy. We further pretrain ViP-CNN on our cleansed Visual Genome Relationship dataset, which is found to perform better than the pretraining on the ImageNet for this task.