A Novel Connectionist System for Unconstrained Handwriting Recognition

  title={A Novel Connectionist System for Unconstrained Handwriting Recognition},
  author={Alex Graves and Marcus Liwicki and Santiago Fern{\'a}ndez and Roman Bertolami and Horst Bunke and J{\"u}rgen Schmidhuber},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
Recognizing lines of unconstrained handwritten text is a challenging task. The difficulty of segmenting cursive or overlapping characters, combined with the need to exploit surrounding context, has led to low recognition rates for even the best current recognizers. Most recent progress in the field has been made either through improved preprocessing or through advances in language modeling. Relatively little work has been done on the basic recognition algorithms. Indeed, most systems rely on… 
Global Context for improving recognition of Online Handwritten Mathematical Expressions
This paper presents a temporal classification method for all three subtasks of symbol segmentation, symbol recognition and relation classification in online handwritten mathematical expressions (HMEs) and shows the effectiveness of the proposed method on the two latest CROHME datasets.
DeepCIN: Attention-Based Cervical histology Image Classification with Sequential Feature Modeling for Pathologist-Level Accuracy
A network pipeline, DeepCIN, to analyze high-resolution epithelium images hierarchically by focusing on localized vertical regions and fusing this local information for determining Normal/CIN classification achieves pathologist-level CIN classification accuracy.
Vehicle License Plate Recognition In Complex Scenes
  • Zhuang Liu, Yuanping Zhu
  • Computer Science
    2020 IEEE 5th International Conference on Intelligent Transportation Engineering (ICITE)
  • 2020
The experimental results show that compared with the existing license plate recognition algorithm, the algorithm in this paper improves significantly the accuracy, and it averages 7.7% in complex scenarios of CCPD dataset.
A Transformer-Based Fusion Recommendation Model For IPTV Applications
This paper presents a deep model for IPTV applications, which generate recommendations using the implicit feedback of users, and a fusion layer is designed on top of the Transformer framework to obtain the semantic preferences of audiences based on their behavior sequences.
What Is Wrong With Scene Text Recognition Model Comparisons? Dataset and Model Analysis
A unified four-stage STR framework is introduced that most existing STR models fit into and allows for the extensive evaluation of previously proposed STR modules and the discovery of previously unexplored module combinations.
Offline Handwriting Recognition on Devanagari Using a New Benchmark Dataset
This paper releases a new handwritten word dataset for Devanagari, IIIT-HW-Dev, and empirically shows that usage of synthetic data and cross lingual transfer learning helps alleviate the issue of lack of training data.
Towards Accurate Handwritten Word Recognition for Hindi and Bangla
This work demonstrates an end-to-end trainable CNN-RNN hybrid architecture which takes inspirations from recent advances of using residual blocks for training convolutional layers, along with the inclusion of spatial transformer layer to learn a model invariant to geometric distortions present in handwriting.
Connectionist Temporal Classification for Offline Handwritten Text Recognition
An overview of the CTC algorithm is given as well as how such algorithm could be speed up using a parallelized implementation.
Context-enhanced text recognition using semantic vector space models
A way to use so called semantic vector space models to add semantic context information to text recognition systems is proposed and experiments show that the proposed methods perform better than N-gram models when trained on the same data.
This work has developed an offline recognition system to recognize the handwritten Devanagari Legal Amount words using neural network, using recurrent neural network classifier as a recognition method.


Long Short-Term Memory
A novel, efficient, gradient based method called long short-term memory (LSTM) is introduced, which can learn to bridge minimal time lags in excess of 1000 discrete-time steps by enforcing constant error flow through constant error carousels within special units.
Token passing: a simple conceptual model for connected speech recognition systems
This paper describes a simple but powerful abstract model in which connected word recognition is viewed as a process of passing tokens around a transition network and some work done under the UK Alvey sponsored VODIS Project is described.
A tutorial on hidden Markov models and selected applications in speech recognition
The fabric comprises a novel type of netting which will have particular utility in screening out mosquitoes and like insects and pests. The fabric is defined of voids having depth as well as width
On-Line and Off-Line Handwriting Recognition: A Comprehensive Survey
The nature of handwritten language, how it is transduced into electronic data, and the basic concepts behind written language recognition algorithms are described.
Offline grammar-based recognition of handwritten sentences
This paper proposes a sequential coupling of a hidden Markov model (HMM) recognizer for offline handwritten English sentences with a probabilistic bottom-up chart parser using stochastic context-free grammars extracted from a text corpus and concludes that syntax analysis helps to improve recognition rates significantly.
Recognition of cursive Roman handwriting: past, present and future
  • H. Bunke
  • Computer Science
    Seventh International Conference on Document Analysis and Recognition, 2003. Proceedings.
  • 2003
The state of the art in off-line Roman cursive handwriting recognition is reviewed, recent trends are analyzed, and challenges for future research in this field are identified.
Handwriting recognition using a reduced character method and neural nets
The reduced character methodology presented in this paper attempts to prove that the recognition of a reduced size character provides a robust approach for recognition of handwritten text.
Connectionist Speech Recognition: A Hybrid Approach
From the Publisher: Connectionist Speech Recognition: A Hybrid Approach describes the theory and implementation of a method to incorporate neural network approaches into state-of-the-art continuous
Probabilistic Interpretation of Feedforward Classification Network Outputs, with Relationships to Statistical Pattern Recognition
  • J. Bridle
  • Computer Science
    NATO Neurocomputing
  • 1989
Two modifications are explained: probability scoring, which is an alternative to squared error minimisation, and a normalised exponential (softmax) multi-input generalisation of the logistic non- linearity of feed-forward non-linear networks with multiple outputs.