Natural Scene Text Understanding

Mancas-Thillou, Céline; Gosselin, Bernard

doi:10.5772/4966

Cited by 24 publications

(3 citation statements)

References 135 publications

(156 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Traditional OCRs are based on carefully crafted features, whereas in modern deep learning-based approaches, informative features are learned from a massive training dataset (e.g., Jaderberg, Vedaldi, & Zisserman, 2014;Zhou et al, 2017). Text detection in video streams, as opposed to text documents or still images, is a challenging task due to several factors, including motion, shadow, lighting, viewing angles, and complex backgrounds, all of which make detecting and extracting text from its background di cult (see Mancas & Gosseli, 2007, for a comprehensive review).…”

Section: Visual Text Indicesmentioning

confidence: 99%

AUVANA: An Automated Video Analysis Tool for Visual Complexity

Alghamdi¹,

Velloso²,

Gruba³

2021

Preprint

View full text Add to dashboard Cite

Visual complexity is widely considered to be an important variable underlying visual perception. While videos have become versatile in their use of visual imagery, surprisingly, little research has been devoted to understanding the impact of visual complexity. In this paper, we present Automated Video Analysis (AUVANA) software, an open-source tool for extracting, computing, and visualizing visual complexity in digital videos. Through leveraging more sophisticated computer vision and video processing algorithms, AUVANA automatically extracts and computes 78 video visual complexity indices. Results of explanatory analyses demonstrated that rather than a unitary construct video visual complexity is more likely a multidimensional and multifaceted phenomenon. We conclude the paper with a discussion about the potential applications of the software

show abstract

Section: Visual Text Indicesmentioning

confidence: 99%

AUVANA: An Automated Video Analysis Tool for Visual Complexity

Alghamdi¹,

Velloso²,

Gruba³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Optical Character Recognition (OCR) can detect and categorize visual patterns into characters from a digital text image [1]. The pattern recognition process includes applying a set of image segmentation, feature extraction, and classification.…”

Section: Introductionmentioning

confidence: 99%

Text Extraction with Optimal Bi-LSTM

H. Nayef,

Norul Huda Sheikh Abdullah,

Sulaiman

et al. 2023

Computers, Materials &Amp; Continua

View full text Add to dashboard Cite

Text extraction from images using the traditional techniques of image collecting, and pattern recognition using machine learning consume time due to the amount of extracted features from the images. Deep Neural Networks introduce effective solutions to extract text features from images using a few techniques and the ability to train large datasets of images with significant results. This study proposes using Dual Maxpooling and concatenating convolution Neural Networks (CNN) layers with the activation functions Relu and the Optimized Leaky Relu (OLRelu). The proposed method works by dividing the word image into slices that contain characters. Then pass them to deep learning layers to extract feature maps and reform the predicted words. Bidirectional Short Memory (BiLSTM) layers extract more compelling features and link the time sequence from forward and backward directions during the training phase. The Connectionist Temporal Classification (CTC) function calcifies the training and validation loss rates. In addition to decoding the extracted feature to reform characters again and linking them according to their time sequence. The proposed model performance is evaluated using training and validation loss errors on the Mjsynth and Integrated Argument Mining Tasks (IAM) datasets. The result of IAM was 2.09% for the average loss errors with the proposed dual Maxpooling and OLRelu. In the Mjsynth dataset, the best validation loss rate shrunk to 2.2% by applying concatenating CNN layers, and Relu.

show abstract

“…The field of character recognition has been studied for a long time and largely used in various fields. The character recognition has been extended to the recognition of characters in natural scene images, which have complicate backgrounds, beyond the stage that recognizes printed letters in simple documents [3] [4]. However, it is difficult to expect a high recognition level as same as that of the recognition in formalized types like documents.…”

Section: Introductionmentioning

confidence: 99%

Correction for Misrecognition of Korean Texts in Signboard Images using Improved Levenshtein Metric

Lee¹

2012

KSII TIIS

View full text Add to dashboard Cite

Recently various studies on various applications using images taken by mobile phone cameras have been actively conducted. This study proposes a correction method for misrecognition of Korean Texts in signboard images using improved Levenshtein metric. The proposed method calculates distances of five recognized candidates and detects the best match texts from signboard text database. For verifying the efficiency of the proposed method, a database dictionary is built using 1.3 million words of nationwide signboard through removing duplicated words. We compared the proposed method to Levenshtein Metric which is one of representative text string comparison algorithms. As a result, the proposed method based on improved Levenshtein metric represents an improvement in recognition rates 31.5% on average compared to that of conventional methods.

show abstract

Natural Scene Text Understanding

Cited by 24 publications

References 135 publications

AUVANA: An Automated Video Analysis Tool for Visual Complexity

AUVANA: An Automated Video Analysis Tool for Visual Complexity

Text Extraction with Optimal Bi-LSTM

Correction for Misrecognition of Korean Texts in Signboard Images using Improved Levenshtein Metric

Contact Info

Product

Resources

About