Accurate Structured-Text Spotting for Arithmetical Exercise Correction

Authors: Yiqing Hu, Yan Zheng, Hao Liu, Dequang Jiang, Yinsong Liu, Bo Ren686-693

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results demonstrate that AEC yields a 93.72% correction accuracy on 40 kinds of mainstream primary arithmetical exercises.
Researcher Affiliation Industry Yiqing Hu, Yan Zheng, Hao Liu, Deqiang Jiang, Yinsong Liu, Bo Ren Youtu Lab, Tencent {hooverhu, neoyzheng, ivanhliu, dqiangjiang, jasonysliu, timren}@tencent.com
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code No The paper states 'We will release this dataset soon' regarding AEC-5k, but does not provide a specific statement or link for the release of source code for the described methodology.
Open Datasets No The paper states, 'We will release these datasets soon.' referring to AEC-5k and the 600k synthetic corpus, but it does not provide a concrete link, DOI, or specific repository name for immediate access to these datasets.
Dataset Splits No The paper mentions '5,000 images for training and 300 images for testing' for AEC-5k, but does not specify a validation dataset split.
Hardware Specification Yes Basing on Pytorch (Paszke et al. 2017), we implement all benchmarks on a regular platform with 8 Nvidia P40 GPUs and 64GB memory.
Software Dependencies No The paper mentions 'Pytorch' but does not specify a version number or other software dependencies with version numbers.
Experiment Setup Yes We adapt the Adam optimizer with learning rate 2.5 10 4 for optimization. We adapt the SGD optimizer with learning rate 0.1 for optimization. The learning rate halves after 300k iterations, and halves again after each 100k iterations.