Harlequin: Color-driven Generation of Synthetic Data for Referring Expression Comprehension
This addresses the data scarcity issue in REC for researchers and practitioners by enabling scalable data generation without manual effort, though it is incremental as it builds on existing data variations.
The paper tackles the problem of expensive manual annotation for Referring Expression Comprehension (REC) by proposing Harlequin, a framework that generates synthetic data with over 1M queries, and shows that pre-training on this artificial data improves model performance on human-annotated datasets.
Referring Expression Comprehension (REC) aims to identify a particular object in a scene by a natural language expression, and is an important topic in visual language understanding. State-of-the-art methods for this task are based on deep learning, which generally requires expensive and manually labeled annotations. Some works tackle the problem with limited-supervision learning or relying on Large Vision and Language Models. However, the development of techniques to synthesize labeled data is overlooked. In this paper, we propose a novel framework that generates artificial data for the REC task, taking into account both textual and visual modalities. At first, our pipeline processes existing data to create variations in the annotations. Then, it generates an image using altered annotations as guidance. The result of this pipeline is a new dataset, called Harlequin, made by more than 1M queries. This approach eliminates manual data collection and annotation, enabling scalability and facilitating arbitrary complexity. We pre-train three REC models on Harlequin, then fine-tuned and evaluated on human-annotated datasets. Our experiments show that the pre-training on artificial data is beneficial for performance.