Strong-Background Restrained Cross Entropy Loss for Scene Text Detection

• Published in: 2019 International Joint Conference on Neural Networks (IJCNN) pp. 1 - 7
• Main Authors: Huang, Randong, Xu, Bo
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.07.2019
• Subjects: Detectors; Entropy; Object detection; Proposals; Testing; Training

In this paper, we investigate the issue of class imbalance in scene text detection. Class Balanced Cross Entropy (CBCE) loss is often adopted for addressing this imbalance problem. We find that CBCE excessively restrains the backward gradients of background. Negative samples own extremely small weig...