You can see below the latest results, updated in real time when new submissions are made to the system. Methods published in ICDAR competition in blue, plus other public methods in white and your methods in yellow.
Description: First, we preprocess the input image with image quality enhancement techniques, and extract character candidates using extremal regions (ER). Second, we verify the extracted character candidates with the character classifier trained by Agile Learning. Afterwards, we do text-patch matching which greatly enhances the recall rate, and group the characters into text regions. Finally, we apply a deep neural network for character recognition.
Description: ASTRoID 1
In the first step our text detection approach uses segmentation algorithm to segment a image into a set of connected components (CCs). The segmentation algorithm is based on the Canny edge detection and dilation methods.
In the second step CCs are described in terms of 49 features and machine learning algorithm (radial SVM) are then used to classify CCs as text or non-text.
In the third step each text CC is further processed in order to detect false positive CCs. This is done by character detection algorithm. Character detection algorithm uses MSER regions which are then described by HOG features and classified by radial SVM and polynomial SVM.
Darko Zelenika (Bosnia and Herzegovina) - Faculty of information studies, Novo mesto, Slovenia
Janez Povh (Slovenia) - Faculty of information studies, Novo mesto, Slovenia
Bernard Ženko (Slovenia) - Jožef Stefan Institute, Ljubljana, Slovenia
Description: This method uses two effective feature vectors for the classification of the text and nontext objects. First feature vector is represented by the Radon transform of text candidate objects. Second feature vector is derived from the detailed geometrical analysis of text contents. Union of two feature vectors is used for the classification of text and nontext objects using support vector machine (SVM). Details of the work can be found at the following publication.
1. Tehsin, S., Masood, A., Kausar, S., & Javed, Y. A CAPTION TEXT DETECTION METHOD FROM IMAGES/VIDEOS FOR EFFICIENT INDEXING AND RETRIEVAL OF MULTIMEDIA DATA. International Journal of Pattern Recognition and Artificial Intelligence. Vol 29 (1), 2015