Please use this identifier to cite or link to this item:
|Title:||Handwriting-OOV word-recognition using web resources||Authors:||Oprean, Cristina
|Affiliations:||Department of Electrical Engineering||Issue Date:||2014||Part of:||La Voisier journal||Volume:||17||Issue:||3||Start page:||77||End page:||96||Abstract:||
Handwriting recognition systems rely on predeﬁned dictionaries. Small and static dictionaries are often exploited to obtain high in-vocabulary (IV) accuracy at the expense of coverage. Thus the recognition of out-of-vocabulary (OOV) words is not handled efﬁciently. To improve OOV recognition while keeping IV dictionaries small, we introduce a multi-step approach that exploits web resources. After an IV-OOV classiﬁcation, Wikipedia is used to create OOV sequence-adapted dynamic dictionaries. A second decoding is done the dynamic dictionary to determine the most probable word for the OOV sequence. We validate our approach with experiments conducted on the RIMES dataset using a BLSTM recognizer. Results show that improvements are obtained compared to handwriting recognition with static dictionary.
|Appears in Collections:||Department of Electrical Engineering|
Show full item record
checked on May 8, 2021
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.