A deep learning based character recognition system from multimedia document

Yadav, Usha; Verma, Satya Bhushan; Xaxa, Deepak Kumar; Mahobiya, Chandrakant

doi:10.1109/ipact.2017.8245200

Cited by 12 publications

(3 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They classified the handwritten and printed samples and discovered that 98.9% of printed data could be identified, whereas only 59.6% of handwritten data could be recognized. CRConvNet [45] and RCNN [46] were used to recognize characters in multimedia documents, while CRConvNet [45] was utilized to detect printed characters. Furthermore, CNN-based OCR was created for the Chinese language [47] and showed promising results.…”

Section: Deep Learning Based Text Recognition Systemsmentioning

confidence: 99%

A Study of Techniques and Challenges in Text Recognition Systems

Kaur,

Kumar

2023

IJRITCC

View full text Add to dashboard Cite

The core system for Natural Language Processing (NLP) and digitalization is Text Recognition. These systems are critical in bridging the gaps in digitization produced by non-editable documents, as well as contributing to finance, health care, machine translation, digital libraries, and a variety of other fields. In addition, as a result of the pandemic, the amount of digital information in the education sector has increased, necessitating the deployment of text recognition systems to deal with it. Text Recognition systems worked on three different categories of text: (a) Machine Printed, (b) Offline Handwritten, and (c) Online Handwritten Texts. The major goal of this research is to examine the process of typewritten text recognition systems. The availability of historical documents and other traditional materials in many types of texts is another major challenge for convergence. Despite the fact that this research examines a variety of languages, the Gurmukhi language receives the most focus. This paper shows an analysis of all prior text recognition algorithms for the Gurmukhi language. In addition, work on degraded texts in various languages is evaluated based on accuracy and F-measure.

show abstract

Section: Deep Learning Based Text Recognition Systemsmentioning

confidence: 99%

A Study of Techniques and Challenges in Text Recognition Systems

Kaur,

Kumar

2023

IJRITCC

View full text Add to dashboard Cite

show abstract

“…Recent works in scene text recognition [32], [2] use convolutional neural networks and pre-processing techniques like adjusting orientations to predict characters from a scene text. Batuhan et al [4] proposed an end-to-end framework of recognizing characters from a document by using LSTMs for character segmentation and CNNs for character classification.…”

Section: Related Workmentioning

confidence: 99%

Fully Convolutional Networks for Handwriting Recognition

Such

Peri

Brockler

et al. 2018

2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR)

View full text Add to dashboard Cite

Handwritten text recognition is challenging because of the virtually infinite ways a human can write the same message. Our fully convolutional handwriting model takes in a handwriting sample of unknown length and outputs an arbitrary stream of symbols. Our dual stream architecture uses both local and global context and mitigates the need for heavy preprocessing steps such as symbol alignment correction as well as complex post processing steps such as connectionist temporal classification, dictionary matching or language models. Using over 100 unique symbols, our model is agnostic to Latin-based languages, and is shown to be quite competitive with state of the art dictionary based methods on the popular IAM and RIMES datasets. When a dictionary is known, we further allow a probabilistic character error rate to correct errant word blocks. Finally, we introduce an attention based mechanism which can automatically target variants of handwriting, such as slant, stroke width, or noise.

show abstract

“…In recent years, the emerging deep learning [13][14][15][16] allows the system to recognize multimedia. Since image, voice, and text usually contain richer information than conventional sensor data, it is promising to use such multimedia data for recognizing fine-grained home contexts.…”

Section: Introductionmentioning

confidence: 99%

Integrating Multiple Models Using Image-as-Documents Approach for Recognizing Fine-Grained Home Contexts

Chen

Saiki

Nakamura

2020

Sensors

View full text Add to dashboard Cite

To implement fine-grained context recognition that is accurate and affordable for general households, we present a novel technique that integrates multiple image-based cognitive APIs and light-weight machine learning. Our key idea is to regard every image as a document by exploiting "tags" derived by multiple APIs. The aim of this paper is to compare API-based models' performance and improve the recognition accuracy by preserving the affordability for general households. We present a novel method for further improving the recognition accuracy based on multiple cognitive APIs and four modules, fork integration, majority voting, score voting, and range voting.Our interest is to use little image data, applying cognitive services to implement affordable context sensing that can adapt to custom contexts in every single house. The cognitive service is a cloud service with cognitive computing functions that provide the capability to understand multimedia data (i.e., vision (object recognition) [19], speech recognition [20], natural language processing [21], etc.), based on sophisticated machine-learning algorithms powered by the offered big data and large-scale computing. They are offered by cloud companies, but the data processing algorithms behind them are not public. They are also widely used in various fields of research, such as modern knowledge management solutions [22] and criminal detection and recognition [23]. The cognitive API is an application program interface via the HTTP/REST protocol [24], with which developers can easily integrate powerful recognition features in their own applications. An image-based cognitive API receives an image from an external application, i.e., extracts specific information from the image, and returns the information in JavaScript Object Notation (JSON) [25] format from the cloud server rather than the local. The information usually contains a set of words called "tags", representing objects and concepts that the API has recognized in the given image. Examples of tags from the API are: [Living, room, indoors, classroom, basement, supporting structure]. The information of interest and the way of recognizing the image vary among individual cognitive services. Related work uses image tagging technology with deep learning, as in [26][27][28], but the implementation is more complex. In our future realistic implementation, for security and privacy of the users the images are not saved after sending the APIs over.The main contribution of this paper is to present a novel method which is not only affordable but also has higher accuracy in recognizing fine-grained home contexts. For this purpose, we are currently investigating techniques that integrate inexpensive camera devices, multiple image-based cognitive APIs, and light-weight machine learning. We previously encoded the tags of a single API to document vectors, then applied them into machine learning for the model construction [29]. However, we found that the accuracy significantly decreased for contexts with multiple people (e.g., "...

show abstract

A deep learning based character recognition system from multimedia document

Cited by 12 publications

References 18 publications

A Study of Techniques and Challenges in Text Recognition Systems

A Study of Techniques and Challenges in Text Recognition Systems

Fully Convolutional Networks for Handwriting Recognition

Integrating Multiple Models Using Image-as-Documents Approach for Recognizing Fine-Grained Home Contexts

Contact Info

Product

Resources

About