Prediction Calibration for Generalized Few-shot Semantic Segmentation
This work addresses a practical but under-studied problem in semantic segmentation for scenarios with limited novel class data, though it appears incremental as it builds on existing GFSS approaches with a novel calibration method.
The paper tackles the problem of base class bias in Generalized Few-shot Semantic Segmentation (GFSS), where existing methods fuse classifier parameters leading to bias toward base classes due to abundant training data. The proposed Prediction Calibration Network (PCN) addresses this by fusing scores from base and novel classifiers using a Transformer-based calibration module, achieving state-of-the-art performance with large margins on PASCAL-5^i and COCO-20^i datasets.
Generalized Few-shot Semantic Segmentation (GFSS) aims to segment each image pixel into either base classes with abundant training examples or novel classes with only a handful of (e.g., 1-5) training images per class. Compared to the widely studied Few-shot Semantic Segmentation FSS, which is limited to segmenting novel classes only, GFSS is much under-studied despite being more practical. Existing approach to GFSS is based on classifier parameter fusion whereby a newly trained novel class classifier and a pre-trained base class classifier are combined to form a new classifier. As the training data is dominated by base classes, this approach is inevitably biased towards the base classes. In this work, we propose a novel Prediction Calibration Network PCN to address this problem. Instead of fusing the classifier parameters, we fuse the scores produced separately by the base and novel classifiers. To ensure that the fused scores are not biased to either the base or novel classes, a new Transformer-based calibration module is introduced. It is known that the lower-level features are useful of detecting edge information in an input image than higher-level features. Thus, we build a cross-attention module that guides the classifier's final prediction using the fused multi-level features. However, transformers are computationally demanding. Crucially, to make the proposed cross-attention module training tractable at the pixel level, this module is designed based on feature-score cross-covariance and episodically trained to be generalizable at inference time. Extensive experiments on PASCAL-$5^{i}$ and COCO-$20^{i}$ show that our PCN outperforms the state-the-the-art alternatives by large margins.