Results 41 to 50 of about 714 (168)

GTC: Guided Training of CTC towards Efficient and Accurate Scene Text Recognition [PDF]

, 2020
Connectionist Temporal Classification (CTC) and attention mechanism are two main approaches used in recent scene text recognition works. Compared with attention-based methods, CTC decoder has a much shorter inference time, yet a lower accuracy. To design
Yi, Shuai, Cai, Xiaocong, Hu, Wenyang, Hou, Jun, Lin, Zhiping   +4 more
core   +1 more source

Research on Continuous Dynamic Gesture Recognition of Chinese Sign Language Based on Multi-Mode Fusion

IEEE Access, 2022
To solve the problem of the low recognition rate of continuous dynamic gestures in Chinese sign language, a non-invasive end-to-end continuous dynamic gesture recognition system combining Inertial Measurement Unit (IMU) signal and surface ...
Jinquan Li, Jiaojiao Meng, Haijun Gong, Zixuan Fan   +3 more
doaj   +1 more source

End to End Alignment Learning of Instructional Videos with Spatiotemporal Hybrid Encoding and Decoding Space Reduction

Applied Sciences, 2021
We solve the problem of how to densely align actions in videos at frame level, with only the order of occurring actions available, in order to save the time-consuming efforts to accurately annotate the temporal boundaries of each action. We propose three
Lin Wang, Xingfu Wang, Ammar Hawbani, Yan Xiong   +3 more
doaj   +1 more source

Pengenalan Karakter Optikal Aksara Jawa Menggunakan Connectionist Temporal Classification [PDF]

Aksara Jawa memiliki sejarah panjang dan penting di Pulau Jawa. Sampai saat ini aksara Jawa banyak digunakan untuk obyek penelitian. Salah satunya menggunakan aplikasi OCR.
Saputro, Wahju Tjahjo, Hidayatullah, Ridwan, Chirzah, Dewi   +2 more
core   +2 more sources

Inter‐Model Feature Fusion for Robust Low‐Resource Speech Recognition

Applied AI Letters, Volume 7, Issue 2, June 2026.
Our Self‐Supervised Feature Fusion (SSF‐FT) method enhances low‐resource speech recognition by adaptively combining features from self‐supervised models trained with Contrastive, Predictive, and Reconstruction objectives. This attention‐weighted ensemble delivers robust performance, particularly in acoustically challenging conditions, extending current
Ussen Kimanuka, Ciira wa Maina, Osman Büyük   +2 more
wiley   +1 more source

Tibetan Data Augmentation via GAN‐Based Handwritten Text Generation

CAAI Transactions on Intelligence Technology, Volume 11, Issue 1, Page 55-65, February 2026.
ABSTRACT Increased awareness of Tibetan cultural preservation, along with technological advancements, has led to significant efforts in academic research on Tibetan. However, the structural complexity of the Tibetan language and limited labeled handwriting data impede advancements in Optical Character Recognition (OCR) and other applications.
Dorje Tashi, Bingtian Chen, Tianying Sheng, Yongbin Yu, Xiangxiang Wang, Jin Zhang, Lobsang Yeshi, Rinchen Dongrub, Thupten Tsering, Nyima Tashi   +9 more
wiley   +1 more source

A deep neural network-based automatic mispronunciation detection in Bengali accented English speech

Discover Computing
Learning a second language, especially English, became necessary as globalisation started. One crucial component of language learning resources is computer-assisted pronunciation training, or CAPT.
Puja Bharati, Sabyasachi Chandra, Aniket Aitawade, Debolina Pramanik, Satya Prasad Gaddamedi, Shyamal Kumar Das Mandal   +5 more
doaj   +1 more source

Bidirectional Representations for Low-Resource Spoken Language Understanding

Applied Sciences, 2023
Speech representation models lack the ability to efficiently store semantic information and require fine tuning to deliver decent performance. In this research, we introduce a transformer encoder–decoder framework with a multiobjective training strategy,
Quentin Meeus, Marie-Francine Moens, Hugo Van hamme   +2 more
doaj   +1 more source

Delay-penalized CTC implemented based on Finite State Transducer [PDF]

, 2023
Connectionist Temporal Classification (CTC) suffers from the latency problem when applied to streaming models. We argue that in CTC lattice, the alignments that can access more future context are preferred during training, thereby leading to higher ...
Kang, Wei, Kuang, Fangjun, Yang, Yifan, Guo, Liyong, Lin, Long, Povey, Daniel, Yang, Xiaoyu, Yao, Zengwei   +7 more
core   +1 more source

Hybrid Deep Learning Frameworks for License Plate Recognition in Complex Environments: A Comprehensive Survey and Research Outlook

IET Computer Vision, Volume 20, Issue 1, January/December 2026.
This survey systematically reviews state‐of‐the‐art license plate recognition methods, with a focus on hybrid CNN‐transformer frameworks and the joint optimisation of detection and recognition for real‐world deployment. It further analyses existing datasets, highlights persistent challenges, such as domain generalisation, and outlines pathways towards ...
SanXing Deng, Ewe Lay Sheng, Tiong Sieh Kiong, MengMeng Yang   +3 more
wiley   +1 more source