Robust ReadingCompetition

method: CLOVA-AI / PAPAGO2018-08-01

Authors: Sunghyo Chung, Youngmin Baek, Hwalsuk Lee, Jaegul Choo

Description: We formulate script identification as a semantic segmentation task. We use both MLT task1 and task2 datasets for training. CLOVA-AI team, Naver Corp.

method: CNN based method 72017-07-02

Authors: Yash Patel, Michal Bušta, Lukáš Neumann, Jiri Matas

Description: A CNN-based approach is used for script- identification in cropped word images. The convolutional lay- ers from VGG-16 architecture are used along with a Global- Average-Pooling and two fully connected layers. To preserve the aspect ratio of input images in both training and testing, the images are resized into fixed-height (64) and variable-width tensors. For training, the convolutional layers are initialized with ImageNet weights. The categorical-cross-entropy loss is utilized, and all the layers (both convolutional and fully connected) are updated during back-propagation.

method: SCUT-DLVClab2017-06-28

Authors: Canjie Luo, Zhaohai Li, Lianwen Jin, Zenghui Sun, Yuliang Liu, Qinghe Zeng

Description: A CNN-based classification method is used. During the training phase, sub-group-of-images are ran- domly cropped. In the test phase, a novel sliding window method is applied on the entire image, which can be regarded as convolutional stride (replaced full connection layer by convolution layer). The category with the top confidence is chosen as the final result. A image-size normalization method is also used for further improving the results.

Ranking Table

Description Paper Source Code
DateMethodScript classification accuracy
2018-08-01CLOVA-AI / PAPAGO89.01%
2017-07-02CNN based method 788.09%
2017-07-01CNN based method 487.33%
2017-07-01CNN based method 586.97%
2017-06-30CNN based method 286.60%
2017-06-02ecn-based method82.20%
2017-07-01An approach towards Word-Level Multi-Script Identification using Deep Transfer Features and SVM74.81%

Ranking Graphic