Text Gestalt: Stroke-Aware Scene Text Image Super-resolution

Authors: Jingye Chen, Haiyang Yu, Jianqi Ma, Bin Li, Xiangyang Xue285-293

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The extensive experimental results validate that the proposed method can indeed generate more distinguishable images on Text Zoom and manually constructed Chinese character dataset Degraded-IC13. Furthermore, since the proposed SFM is only used to provide stroke-level guidance when training, it will not bring any time overhead during the test phase.
Researcher Affiliation Academia 1Shanghai Key Laboratory of Intelligent Information Processing School of Computer Science, Fudan University 2The Hong Kong Polytechnic University
Pseudocode No The paper describes the method verbally but does not include structured pseudocode or algorithm blocks.
Open Source Code Yes Code is available at https://github.com/FudanVI/FudanOCR.
Open Datasets Yes The datasets used in this paper are as follows: Text Zoom (Wang et al. 2020) The images in Text Zoom originate from Real SR (Cai et al. 2019) and SR-RAW (Zhang et al. 2019). ...IC15 (Karatzas et al. 2015)... Degraded-IC13 is constructed based on IC13-HCCR (Yin et al. 2013)... After pretrained on Synth90k (Jaderberg et al. 2016) and Synth Text (Gupta, Vedaldi, and Zisserman 2016)...
Dataset Splits Yes Text Zoom contains 17, 367 LR-HR pairs for training and 4, 373 pairs for testing. In terms of different focal lengths of digital cameras, the test set is divided into three subsets, including 1, 619 LR-HR pairs for the easy subset, 1, 411 LR-HR pairs for the medium subset, and 1, 343 LR-HR pairs for the hard subset. ... We employ CRNN for validation.
Hardware Specification Yes All experiments are conducted on one NVIDIA GTX 1080Ti GPU with 11GB memory.
Software Dependencies No The paper mentions "Our model is implemented in Py Torch" but does not specify a version number for PyTorch or any other software dependencies with their versions.
Experiment Setup Yes The model is trained using Adam (Kingma and Ba 2014) optimizer with learning rate set to 10 4. The batch size is set to 16.