Deep Learning for Fast and Accurate Horizontal English Text Detection

Student: Zalesskaia Galina

Supervisor: Alexandr Rassadin

Faculty: Faculty of Informatics, Mathematics, and Computer Science (HSE Nizhny Novgorod)

Educational Programme: Applied Mathematics and Information Science (Bachelor)

Final Grade: 9

Year of Graduation: 2020

This investigation focuses on the text detection task. The main goal is to build a fast and accurate text detector that achieves real-time speed without using GPUs. To accomplish this task, its area of application was narrowed to the detection of horizontal English text. During the experiments, a light neural network with 88.17 F-score and 41.2 FPS on CPU was obtained. In addition to horizontal text, it finds even small, handwritten, rotated or unfocused text. It also detects words in French, German and other Latin-based languages. The model with configuration files and scripts has been added to the "OpenVINO™ Training Extensions" repository and is available to any user.

Full text (added May 25, 2020)

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses