Cycle-Consistency Learning for Captioning and Grounding
This work addresses the challenge of integrating visual grounding and captioning for researchers in computer vision, offering a novel collaborative training approach that is incremental in combining existing tasks.
The paper tackles the problem of independent training pipelines for visual grounding and image captioning by proposing CyCo, a cyclic-consistent learning framework that bridges these tasks. The result includes state-of-the-art performance in fully supervised visual grounding, competitive semi-weakly supervised grounding, and a captioning model capable of describing arbitrary image regions with strong benchmark results.
We present that visual grounding and image captioning, which perform as two mutually inverse processes, can be bridged together for collaborative training by careful designs. By consolidating this idea, we introduce CyCo, a cyclic-consistent learning framework to ameliorate the independent training pipelines of visual grounding and image captioning. The proposed framework (1) allows the semi-weakly supervised training of visual grounding; (2) improves the performance of fully supervised visual grounding; (3) yields a general captioning model that can describe arbitrary image regions. Extensive experiments show that our fully supervised grounding model achieves state-of-the-art performance, and the semi-weakly supervised one also exhibits competitive performance compared to the fully supervised counterparts. Our image captioning model has the capability to freely describe image regions and meanwhile shows impressive performance on prevalent captioning benchmarks.