REALEDIT: Reddit Edits As a Large-scale Empirical Dataset for Image Transformations
This addresses the lack of realistic training data for image editing models, benefiting users and developers, though it is incremental as it focuses on dataset creation and model adaptation.
The authors tackled the problem of image editing models failing to meet real-world user needs by introducing REALEDIT, a large-scale dataset of authentic user requests and human-made edits from Reddit, and their model achieved up to 165 Elo points improvement in human judgment and a 92% relative gain on an automated metric.
Existing image editing models struggle to meet real-world demands. Despite excelling in academic benchmarks, they have yet to be widely adopted for real user needs. Datasets that power these models use artificial edits, lacking the scale and ecological validity necessary to address the true diversity of user requests. We introduce REALEDIT, a large-scale image editing dataset with authentic user requests and human-made edits sourced from Reddit. REALEDIT includes a test set of 9300 examples to evaluate models on real user requests. Our results show that existing models fall short on these tasks, highlighting the need for realistic training data. To address this, we introduce 48K training examples and train our REALEDIT model, achieving substantial gains - outperforming competitors by up to 165 Elo points in human judgment and 92 percent relative improvement on the automated VIEScore metric. We deploy our model on Reddit, testing it on new requests, and receive positive feedback. Beyond image editing, we explore REALEDIT's potential in detecting edited images by partnering with a deepfake detection non-profit. Finetuning their model on REALEDIT data improves its F1-score by 14 percentage points, underscoring the dataset's value for broad applications.