EnsNet: Ensconce Text in the Wild

Authors: Shuaitao Zhang, Yuliang Liu, Lianwen Jin, Yaoxiong Huang, Songxuan Lai801-808

AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Both qualitative and quantitative sensitivity experiments on synthetic images and the ICDAR 2013 dataset demonstrate that each component of the Ens Net is essential to achieve a good performance. Moreover, our Ens Net can significantly outperform previous state-of-the-art methods in terms of all metrics.
Researcher Affiliation Academia Shuaitao Zhang, Yuliang Liu, Lianwen Jin, Yaoxiong Huang, Songxuan Lai School of Electronic and Information Engineering South China University of Technology
Pseudocode No The paper describes the methodology using prose and diagrams (Figure 2, Figure 3) but does not include any explicit pseudocode or algorithm blocks.
Open Source Code Yes The synthetic data are available at: https://github.com/HCIILAB/Scene-Text-Removal.
Open Datasets Yes However, such text dataset does not exist currently; thus, we constructed a synthetic dataset for evaluating the performance. In addition, we evaluated the performance on the ICDAR 2013 (Karatzas et al. 2013) dataset... The synthetic data are available at: https://github.com/HCIILAB/Scene-Text-Removal.
Dataset Splits No In our experiments, the training set consists of a total of 8000 images and the test set contains 800 images; all the training and test samples are resized to 512 × 512.
Hardware Specification Yes Ens Net is extremely fast, which can preform at 333 fps on an i5-8600 CPU device.
Software Dependencies No The paper does not provide specific version numbers for any software dependencies or libraries used in the experiments.
Experiment Setup Yes In our experiments, λe, λi, and λt are empirically set to 0.5, 50.0, and 25.0, respectively. All experiments use exactly the same settings (input size is set to 512 × 512).