Visual Question Generation as Dual Task of Visual Question Answering
This work addresses the challenge of integrating VQA and VQG for better performance in computer vision applications, though it is incremental as it builds on existing VQA architectures.
The authors tackled the problem of visual question answering (VQA) and visual question generation (VQG) by proposing an end-to-end unified framework called iQAN, which jointly trains on both tasks to leverage their complementary relations, resulting in improved VQA accuracy on datasets like CLEVR and VQA2 over baselines.
Recently visual question answering (VQA) and visual question generation (VQG) are two trending topics in the computer vision, which have been explored separately. In this work, we propose an end-to-end unified framework, the Invertible Question Answering Network (iQAN), to leverage the complementary relations between questions and answers in images by jointly training the model on VQA and VQG tasks. Corresponding parameter sharing scheme and regular terms are proposed as constraints to explicitly leverage Q,A's dependencies to guide the training process. After training, iQAN can take either question or answer as input, then output the counterpart. Evaluated on the large-scale visual question answering datasets CLEVR and VQA2, our iQAN improves the VQA accuracy over the baselines. We also show the dual learning framework of iQAN can be generalized to other VQA architectures and consistently improve the results over both the VQA and VQG tasks.