Correction: DC-PSENet: a novel scene text detection method integrating double ResNet-based and changed channels recursive feature pyramid

Correction to: The Visual Computer (2024) 40:4473–4491https://doi.org/10.1007/s00371-023-03093-5

The publication of this article unfortunately contained mistakes. There were errors in table 1 and table 5. The corrected tables are given below.

1. Incorrect Table 1:

Step	Meaning	Value
Set learning rate about training	Learning rate	0.001
Set the number of iterations during training	Number of iterations	600
Set the loss starting value	Loss text value	1
Set the value to balance between loss text and loss kernel	Balance value μ	0.7
Set the value to balance between ResNet50-Mish and Res2Net50-Mish	Weights η	0.5(ICDAR15) or 0.4(CTW1500)

Table 1 is corrected as follows:

Step	Meaning	Value
Set learning rate about training	Learning rate	0.001
Set the number of iterations during training	Number of iterations	600 × 250
Set the loss starting value	Loss text value	1
Set the value to balance between loss text and loss kernel	Balance value μ	0.7
Set the value to balance between ResNet50-Mish and Res2Net50-Mish	Weights η	0.5(ICDAR15) or 0.4(CTW1500)

2. Incorrect Table 5:

Item	Backbone		Neck
Time(s)	ResNet50	DRN	FPN	CCRFP
Time(ICDAR15)	0.0430	0.1375	0.0138	0.0299
Time(CTW1500)	0.0358	0.1106	0.0140	0.0246

Table 5 is corrected as follows:

Item	Backbone		Neck
Time(s)	ResNet50	DRN	FPN	CCRFP
Time(ICDAR15)	0.0430	0.1375	0.0138	0.0299
Time(CTW1500)	0.0358	0.1106	0.0109	0.0246

The original article has been corrected.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.