Annotating Object Instances with a Polygon-RNN
This addresses the time-consuming and costly process of manual object segmentation for dataset creation, offering a practical tool for computer vision researchers and annotators.
The paper tackles the problem of semi-automatic object instance annotation by predicting polygons sequentially, which speeds up annotation by 4.7x across Cityscapes classes with 78.4% IoU agreement, matching human annotator consistency.
We propose an approach for semi-automatic annotation of object instances. While most current methods treat object segmentation as a pixel-labeling problem, we here cast it as a polygon prediction task, mimicking how most current datasets have been annotated. In particular, our approach takes as input an image crop and sequentially produces vertices of the polygon outlining the object. This allows a human annotator to interfere at any time and correct a vertex if needed, producing as accurate segmentation as desired by the annotator. We show that our approach speeds up the annotation process by a factor of 4.7 across all classes in Cityscapes, while achieving 78.4% agreement in IoU with original ground-truth, matching the typical agreement between human annotators. For cars, our speed-up factor is 7.3 for an agreement of 82.2%. We further show generalization capabilities of our approach to unseen datasets.