Deep Learning for Earth Image Segmentation based on Imperfect Polyline Labels with Annotation Errors
This addresses a critical issue for researchers and practitioners in remote sensing and earth observation, where imperfect labels degrade segmentation performance, though it is an incremental improvement over existing error-handling methods.
The paper tackles the problem of geometric annotation errors in ground truth labels for earth image segmentation by proposing a learning framework that simultaneously updates model parameters and infers true label locations, resulting in a 67% reduction in false positives and 55% reduction in false negatives on a hydrological dataset.
In recent years, deep learning techniques (e.g., U-Net, DeepLab) have achieved tremendous success in image segmentation. The performance of these models heavily relies on high-quality ground truth segment labels. Unfortunately, in many real-world problems, ground truth segment labels often have geometric annotation errors due to manual annotation mistakes, GPS errors, or visually interpreting background imagery at a coarse resolution. Such location errors will significantly impact the training performance of existing deep learning algorithms. Existing research on label errors either models ground truth errors in label semantics (assuming label locations to be correct) or models label location errors with simple square patch shifting. These methods cannot fully incorporate the geometric properties of label location errors. To fill the gap, this paper proposes a generic learning framework based on the EM algorithm to update deep learning model parameters and infer hidden true label locations simultaneously. Evaluations on a real-world hydrological dataset in the streamline refinement application show that the proposed framework outperforms baseline methods in classification accuracy (reducing the number of false positives by 67% and reducing the number of false negatives by 55%).