Rosetta: Large Scale System for Text Detection and Recognition in Images
Open accessThe DOI matches its Crossref record; 31 of the 31 references it cites were found in a registry or an index.
Abstract
In this paper we present a deployed, scalable optical character recognition (OCR) system, which we call Rosetta , designed to process images uploaded daily at Facebook scale. Sharing of image content has become one of the primary ways to communicate information among internet users within social networks such as Facebook, and the understanding of such media, including its textual information, is of paramount importance to facilitate search and recommendation applications. We present modeling techniques for efficient detection and recognition of text in images and describe Rosetta 's system architecture. We perform extensive evaluation of presented technologies, explain useful practical approaches to build an OCR system at scale, and provide insightful intuitions as to why and how certain components work based on the lessons learnt during the development and deployment of the system.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“A novel OCR system is developed, named as Rosetta, wherein faster-RCNN is utilized for detecting the text characters from millions of Facebook images, however, fully convolutional CNN is employed for generating the lexicon free transcription of each word [6].”
From Object detection using YOLO: challenges, architectural successors, datasets and applications (2023) · cites this paper as [6]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 3×
“chitectural decision and a natural choice was to use a character-based recognition model. In the initial stages of development, we tested various efficient backbone architectures, including ResNet-18 [17] and SqueezeNet [18]. We evaluated these choices on our synthetic data and ended up selecting ResNet-18 as our main…”
MethodInfluentialFaster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Ren, S., He, K., Girshick, R., & Sun, J. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2577031 · Cited by 73,472 · cited 3×
“Our text detection approach, based on Faster-RCNN model [24], is responsible of detecting regions of the image that contain text.”
MethodInfluentialImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333 · cited 1×
“ns. As we show empirically in Section 5, ShuffleNet is significantly faster than ResNet and did not lead to inferior accuracy. The ShuffleNet convolutional trunk is pre-trained using ImageNet dataset [10]. To train the end-to-end detection system, the model was bootstrapped with an in-house synthetic dataset and then…”
MethodShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices
Zhang, X., Zhou, X., Lin, M., & Sun, J. (2017) · arXiv · doi:10.48550/arxiv.1707.01083 · Cited by 9,982 · cited 1×
“e encoding, RPN, and classifiers) are trained jointly in a supervised, end-to-end manner. Our text detection model uses Faster-RCNN, but replaces the ResNet convolutional body with a ShuffleNet-based [31] architecture for efficiency reasons. As we show empirically in Section 5, ShuffleNet is significantly faster than…”
MethodShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices
Zhang, X., Zhou, X., Lin, M., & Sun, J. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00716 · Cited by 9,982
SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size
Iandola, F. N., Han, S., Moskewicz, M. W., Ashraf, K., Dally, W. J., & Keutzer, K. (2016) · arXiv · doi:10.48550/arxiv.1602.07360 · Cited by 6,201 · cited 2×
“st of the k fully connected layers that perform character classification in the CHAR model. Orthogonal to this, one can use different body architectures. We experimented with ResNet-18 and SqueezeNet [18], and obtained higher accuracy with ResNet-18 while experiencing only a very small overhead in computation. Image P…”
MethodGraves, A., Fernández, S., Gomez, F., & Schmidhuber, J. (2006) · Proceedings of the 23rd international conference on Machine learning - ICML '06 · doi:10.1145/1143844.1143891 · Cited by 5,370
Bengio, Y., Louradour, J., Collobert, R., & Weston, J. (2009) · Proceedings of the 26th Annual International Conference on Machine Learning · doi:10.1145/1553374.1553380 · Cited by 5,366 · cited 1×
“The second approach was based on curriculum learning [6], i.e ., starting with a simpler problem and increasing the difficulty as the model improves.”
BackgroundShi, B., Bai, X., & Yao, C. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2646371 · Cited by 3,172
Edge Boxes: Locating Object Proposals from Edges
Zitnick, C. L., & Dollár, P. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10602-1_26 · Cited by 2,656 · cited 1×
“Before the recognition stage, a combination of Edge Boxes [33] and an aggregate channel features (ACF) detector framework [11] is used to detect candidate locations for the words.”
Method
Research works printed but not linked: 1
- [2] Caffe2, 2017. -
Cited by
Items 301–400 of 425
Research on Chinese License Plate Automatic Recognition Based on Transfer Learning
Cai, X., Liao, G., & Zhao, H. (2024) · International Conference on Cyber Security and Information Engineering (ICCSIE) · doi:10.1145/3689236.3689242 · Cited by 1
“By transferring knowledge from large [ 3 ] , general datasets to specific tasks, transfer learning enables the training of accurate models even with limited target-domain data .”
Compressed Vision Transformer for Scene Text Recognition
Ren, J., Deng, T., Huang, Y., Qu, D., Su, J., & Li, B. (2024) · International Conference on Algorithms, Computing and Artificial Intelligence (ACAI) · doi:10.1109/acai63924.2024.10899477 · Cited by 1
“( 1 × 10 6 ) ( 1 × 10 9 ) CRNN [17] 76.7 8.5 1.4 R2AM [18] 78.4 2.9 2.0 GRCNN [19] 78.3 11.2 1.8 Rosetta [20] 78.4 5.3 10.1 RARE [21] 82.1 10.8 2.0 STAR-Net [22] 81.8 48.9 10.7 ViTSTR-Tiny [14] 80.”
InfluentialBilingual Road Text Recognition Based on a Hybrid Model of CTC and Attention
Zaghdoud, R., Boukthir, K., Hamdani, T. M., & Alimi, A. M. (2024) · IEEE/ACS International Conference on Computer Systems and Applications (AICCSA) · doi:10.1109/aiccsa63423.2024.10912568 · Cited by 1
A real-time text recognition based on involution operators and graph convolutional networks
Tong, G., Li, Y., Dong, M., & Peng, D. (2025) · Journal of Real-Time Image Processing · doi:10.1007/s11554-025-01679-x · Cited by 1
“For example, Rosetta [1] used features extracted from convolutional neural networks to predict the feature sequences using the ResNet model as a backbone.”
Khan, D. (2025) · INTERANTIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING AND MANAGEMENT · doi:10.55041/ijsrem46510 · Cited by 1
A Multi-Modal Attentive Framework That Can Interpret Text (MMAT)
Kumari, V., Gupta, S., Sharma, Y., & Goel, L. (2025) · IEEE Access · doi:10.1109/access.2025.3586900 · Cited by 1
“Some best-performing OCR systems, such as Google Vision API [39], and Facebook’s Rosetta [40], fail to perform very well in real-time constraints.”
Raisi, Z., Nazarzehi, V., Sarani, E., & Damani, R. (2025) · SN Computer Science · doi:10.1007/s42979-025-04196-7 · Cited by 1
AI-Powered Texts Extracted from Images To Speech Synthesis
Boumedjout, A., Kabli, F., Bendimrad, N., Kebdani, S., & Saidi, B. (2025) · Advances in intelligent systems research/Advances in Intelligent Systems Research · doi:10.2991/978-94-6463-805-9_25 · Cited by 1
Wang, H., & Zhang, D. (2025) · IEEE International Conference on Automation Science and Engineering (CASE) · doi:10.1109/case58245.2025.11164100 · Cited by 1
“These findings validate the effectiveness of GCN correction 0.973 Rosetta [15] 0.853 Rosetta [15] + GCN(0.7) 0.886 CRNN [16] 0.868 CRNN [16] + GCN(0.7) 0.903 STAR-Net [17] 0.887 STAR-Net [17] + GCN(0.7) 0.912 in addressing complex sequence dependencies and layout rules in elevator panel images.”
InfluentialFarshidi, K., Khotanlou, H., & Alighardash, E. (2026) · Multimedia Tools and Applications · doi:10.1007/s11042-026-21656-1 · Cited by 1
“4 LoRRA model [2] f V Text reading (OCR) component An OCR system [53] is used to extract text from the image.”
Yi, Z., Zhao, J., Lv, J., & Wang, T. (2026) · Information Fusion · doi:10.1016/j.inffus.2026.104700 · Cited by 1
Managing scalable direct storage accesses for GPUs
Zhou, Y. (. (2025) · Illinois Digital Environment for Access to Learning and Scholarship (University of Illinois at Urbana-Champaign) · Cited by 1
Document Data Extraction System Based on Visual Words Codebook
Loginov, V. N., Valiukov, A., Semenov, S., & Zagaynov, I. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-57058-3_25 · Cited by 1
Toward an End-to-End Solution to Identification of Handheld Pharmaceutical Blister Packages
Chung, S., Cho, C.-L., & Su, S. (2020) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc42975.2020.9283025 · Cited by 1
“Facebook team proposes Rosetta [6], which uses Faster R-CNN [2] for text detection and fully-convolution with CTC loss [7] for text recognition.”
MethodMulti-granularity Deep Local Representations for Irregular Scene Text Recognition
Gao, H., Li, Y., Dai, J., Wang, X., Han, J., & Li, R. (2021) · ACM/IMS Transactions on Data Science · doi:10.1145/3446971 · Cited by 1
Orihashi, S., Yamazaki, Y., Makishima, N., Ihori, M., Takashima, A., Tanaka, T., & Masumura, R. (2021) · ACM International Conference on Multimedia in Asia (MMAsia) · doi:10.1145/3469877.3490571 · Cited by 1
“First, the recognition models based on Transformer have the same or higher Japanese character recognition accuracy than the recognition model based on RNNs with limited training data.”
BackgroundMethodInfluentialDevelopment of Digital Storytelling Platform for Children based on Emotions
Vijayakumaran, T., Thavachelvam, T., Gnaeswaran, A., Sandanayake, T., Sumathipala, K., & Bandara, S. (2021) · International Conference on Information Technology Research (ICITR) · doi:10.1109/icitr54349.2021.9657431 · Cited by 1
Wu, Q., Wang, P., Wang, X. Y., He, X., & Zhu, W. (2022) · Advances in computer vision and pattern recognition · doi:10.1007/978-981-19-0964-1_12 · Cited by 1
Improved Telugu Scene Text Recognition with Thin Plate Spline Transform
Rao, N. S., & Negi, A. (2022) · Lecture notes on data engineering and communications technologies · doi:10.1007/978-981-16-9113-3_65 · Cited by 1
Krokida, Z. (2022) · De Gruyter eBooks · doi:10.1515/9783110732009-019 · Cited by 1
Rwin-FPN++: Rwin Transformer with Feature Pyramid Network for Dense Scene Text Spotting
Zeng, C., Liu, Y. F., & Song, C. (2022) · Applied Sciences · doi:10.3390/app12178488 · Cited by 1
AccNet: occluded scene text enhancing network with accretion blocks
Gong, Y., Zhang, Z., Duan, G., Ma, Z., & Xie, M. (2022) · Machine Vision and Applications · doi:10.1007/s00138-022-01351-5 · Cited by 1
“Afterward, some methods tried to utilize Residual Networks (ResNet) [19] for better feature extraction [9,20], or involve attention mechanism to make further progresses [8,21].”
MethodKnowledge-Based Scene Text Recognition for Industrial Applications
Deng, G., Tu, J., Chen, C., He, J., & Le, X. (2022) · IEEE International Conference on Industrial Technology (ICIT) · doi:10.1109/icit48603.2022.10002724 · Cited by 1
Text Spotting in Natural Scenes Based on Feature Pyramid Neural Network
Huang, C., & Liou, G. (2022) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata55660.2022.10020609 · Cited by 1
Omar, I. O., Haboubi, S., & Benzarti, F. (2022) · IEEE Information Technologies & Smart Industrial Systems (ITSIS) · doi:10.1109/itsis56166.2022.10118367 · Cited by 1
“were able to localize text blocks within the input image [4][15].”
BackgroundMethodInfluentialLCSTR: Scene Text Recognition with Large Convolutional Kernels
Wang, J., Yang, L., Wang, J., Yang, H., Bai, L., Wang, P. S. P., Li, X., Luo, H., & Xu, H. (2023) · International Journal of Pattern Recognition and Artificial Intelligence · doi:10.1142/s021800142353004x · Cited by 1
Emotional Storyteller for Vision Impaired and Hearing-Impaired Children
Kengatharan, E. A. A. (2023) · International Journal on Recent and Innovation Trends in Computing and Communication · doi:10.17762/ijritcc.v11i10.8562 · Cited by 1
STUDY OF MACHINE LEARNING TOOLS AND ALGORITHMS FOR RECOGNITION AND DIGITALISATION OF SALES RECEIPTS
Kandyba, V., Kushnir, O., Bredikhin, V., & Khoroshylova, I. (2023) · Municipal economy of cities · doi:10.33042/2522-1809-2023-6-180-7-11 · Cited by 1
Two-Stage Reasoning Network with Modality Decomposition for Text VQA
Ling, S., You, S., & Bao, B. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-53311-2_10 · Cited by 1
Multi-modal Domain Adaptation for Text Visual Question Answering Tasks
Li, Z., Liu, D., & Cai, W. (2023) · International Conference on Digital Image Computing: Techniques and Applications (DICTA) · doi:10.1109/dicta60407.2023.00058 · Cited by 1
“The Rosetta model [37] is chosen as an OCR detector to explore scene information including visual objects, coordinates, FastText [28], and PHOC [29] features.”
MethodAdio, A. O., Akanbi, C. O., Adigun, A. A., Kasali, A. A., & Adisa, S. (2024) · Dutse Journal of Pure and Applied Sciences · doi:10.4314/dujopas.v10i2b.17 · Cited by 1
“Borisyuk et al, (2018) suggested developing a reliable and precise optical character recognition system that could process hundreds of millions of photos every day in real time.”
BackgroundDocLightDetect: A New Algorithm for Occlusion Classification in Identification Documents
das Neves, R. B., Bezerra, B. L. D., & Zanchettin, C. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-70442-0_12 · Cited by 1
Text Visual Question Answering Based on Interactive Learning and Relationship Modeling
Zhang, C., Wu, W., & Ma, B. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72347-6_7 · Cited by 1
Han, S., Zhou, J. L., Zhu, X., Li, Z., Liu, J., Wang, H., & Gong, Y. (2022) · Lecture notes in computer science · doi:10.1007/978-981-96-8298-0_4 · Cited by 1
Fine-Grained WEEE Product Retrieval Using Hybrid Search and Re-ranking
Obayemi, A., & Nguyen, K. A. (2025) · Lecture notes in computer science · doi:10.1007/978-981-96-8892-0_32 · Cited by 1
Dsrn-svmamba: a dual-stream recursive network base on SVMamba for scene text recognition
Zhao, Y., Ding, F., Zhang, D., Yang, Z., & Jiang, J. (2025) · The Journal of Supercomputing · doi:10.1007/s11227-025-07888-2 · Cited by 1
Using Computer Vision to Identify Objects in an Operating Theatre to Support Safer Working
Stephen, O. (2023) · Tuwhera (Auckland University of Technology) · Cited by 1
Facebook Spam Detection Extension Tool for Chromium Browser Architecture
Azman, L. H., & Abdullah, N. A. (2023) · Applied Information Technology And Computer Science · Cited by 1
A Machine Learning Framework for Data Ingestion in Document ImagesPreprint
Fu, H., Bai, Y., Li, Z., Shen, J., & Sun, J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.00838 · Cited by 1
“Object detection is a hot research topic in computer vision area and state-of-art approaches [7, 9, 24] have been utilized for words detection [3, 33, 35].”
BackgroundMethodVisual Recognition Based System To Assist Blind Persons
Dongre, A. (2020) · TRAP@NCI (National College of Ireland) · Cited by 1
Using Human Psychophysics to Evaluate Generalization in Scene Text Recognition ModelsPreprint
Siddiqui, S., Sizikova, E., Roig, G., Majaj, N. J., & Pelli, D. G. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.00083 · Cited by 1
“Alternatively, attention-based methods [7, 33, 17, 6, 21] learn a character-level language model, and predict each output character by attending only to a certain predicted region of the input.”
MethodEASTER: Efficient and Scalable Text RecognizerPreprint
Chaudhary, K., & Bali, R. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2008.07839 · Cited by 1
Structured Models for Vision-and-Language Reasoning
Hu, R. (2020) · eScholarship (California Digital Library) · Cited by 1
Epidemiology Inspired Framework for False Information Mitigation in Social Networks
Rath, B. (2020) · University of Minnesota Digital Conservancy (University of Minnesota) · Cited by 1
Text-Aware Dual Routing Network for Visual Question AnsweringPreprint
Jiang, L., He, Y., & Chen, J. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2211.14450 · Cited by 1
“Given an image with M OCR tokens derived from the Rosetta system (Borisyuk, Gordo, and Sivakumar 2018), we introduce the OCR feature for each OCR patch following the setting in (Hu et al. 2020).”
MethodAdvanced document data extraction techniques to improve supply chain performance
Sharma, V. (2021) · Repository@Hull (Worktribe) (University of Hull) · Cited by 1
Segmentation, Recognition and Indexing of Cham characters in Cham documents
Nguyen, T.-N. (2023) · theses.fr (ABES) · doi:10.70675/372403a2z0923z4315z8ff6z716159bb334f · Cited by 1
Al-Azzawi, S., Liu, C., Habib, N., Barney, E., & Liwicki, M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.05900 · Cited by 1
“Arabic Persian Urdu ﺍ ﺏ ﺕ ﺙ ﺝ ﺡ ﺥ ﺩ ﺫ ﺭ ﺯ ﺵ ﺱ ﺹ and memory consumption [3, 4].”
An Approach for Detecting Image Spam in OSNs
Imam, N. H., & Vassilakis, V. G. (2019) · Conference for Truth and Trust Online (TTO) · doi:10.36370/tto.2019.10
“Facebook, the most popular OSN in the world, had 1.87 billion monthly active users and 1.15 billion daily active users during the period of January to February 2017 (Watcharenwong and Saikaew, 2017).”
BackgroundMethodInfluentialUnsupervised preprocessing of medical imaging data with generative adversarial networks
Kirkerød, M. (2019) · NORA - Norwegian Open Research Archives
Flexible scene text recognition based on dual attention mechanism
Tian, Z., Wang, C., Xiao, Y., & Lin, Y. (2020) · Concurrency and Computation Practice and Experience · doi:10.1002/cpe.5863
Dual Path Multi-Modal High-Order Features for Textual Content based Visual Question Answering
Li, Y., Lin, Y., Zhao, H., & Wang, D. (2021) · Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition · doi:10.1109/icpr48806.2021.9412231
“It also provides OCR information of each image recognized by Rosetta system [43] to help us focusing more on the model itself.”
BackgroundDetection of Text from Video with Customized Trained Anatomy
Mortha, M. D. D., Maddala, S., & Raju, V. (2021) · International Conference on Data Science, E-Learning and Information Systems (DATA) · doi:10.1145/3460620.3460623
“Sample dataset can be seen in figure 1 [17].”
BackgroundMethodSeo, M., Kim, D., & Choi, D. (2021) · The Journal of Korea Robotics Society · doi:10.7746/jkros.2021.16.3.276
Key-Guided Identity Document Classification Method by Graph Attention Network
Xia, X., Liu, W., Zhang, Y., Wang, L., & Sun, J. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-86337-1_18
A Method for Identifying the Key Information of Electronic Invoicing in Complex Scenes
Zhi, X., Shen, Z., & Zhao, B. (2021) · International Conference on Image, Vision and Computing (ICIVC) · doi:10.1109/icivc52351.2021.9526973
Fast Arbitrary Shaped Scene Text Detection via Text Discriminator
Zeng, C., & Song, C. (2021) · Journal of Physics Conference Series · doi:10.1088/1742-6596/2025/1/012014
“Borisyuk et al. [5] adopted famous object detection network Faster-RCNN [6] for text detection, and a sequence-to-sequence CTC loss [7] for training text recognition.”
DSAMT: Dual-Source Aligned Multimodal Transformers for TextCaps
Liao, C., Liu, R., & Gao, S. (2021) · IEEE International Conference on Network Infrastructure and Digital Content (IC-NIDC) · doi:10.1109/ic-nidc54101.2021.9660575
Adisaisakda, P., Tiawongsombat, P., & Laotaweesub, W. (2022) · The Journal of King Mongkut s University of Technology North Bangkok · doi:10.14416/j.kmutnb.2022.02.007
Research on the Application of Speech Recognition in Medical Clinic
Jin, H. (2021) · International Conference on Computer Information Science and Artificial Intelligence (CISAI) · doi:10.1109/cisai54367.2021.00010
Cut To Size: A Lightweight Design for STR via Channel Pruning
Zhang, N., Li, C., Chang, E., Liu, F., Xuan, S., & Chen, M. (2021) · China Automation Congress (CAC) · doi:10.1109/cac53003.2021.9727805
Relation-Aware Global-Augmented Transformer for TextCaps
Li, Q., Li, B., & Ma, C. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-15919-0_62
Smart Text Reader System for People who are Blind Using Machine and Deep Learning
Raisi, Z., Naiel, M. A., Younes, G., Fieguth, P., & Zelek, J. (2022) · Wiley eBooks · doi:10.1002/9781119861850.ch10
Cai, G., Qu, L., Li, Y., Cheng, G., Lu, X., Wang, Y., Yao, F., & Wang, S. (2022) · International Journal of Digital Crime and Forensics · doi:10.4018/ijdcf.315745
“, 2018), Rosetta (Borisyuk et al., 2018), and CRNN (Shi et al.”
BackgroundMethodLin, J., Guo, M., Gong, L., & Hu, J. (2022) · International Conference on Information Economy, Data Modeling and Cloud Computing (ICIDC) · doi:10.4108/eai.17-6-2022.2322692
“Chinese natural language interface and SQL have very close expression ability, but different from SQL, it is natural for users and easier to master [5].”
BackgroundA Scheduled Mask Method for TextVQA
Han, M., Jin, T., & Lin, W. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-23585-6_3
A Framework for Handwritten Date Recognition in Quality Documents
Tang, H., Zhang, J., Yan, K., Wang, G., & Wang, H. (2022) · IEEE International Conference on e-Business Engineering (ICEBE) · doi:10.1109/icebe55470.2022.00048
“Rosetta [10] consists of a fully convolutional model and CTC, which also achieves great consequences.”
BackgroundAre Attention blocks better than BiLSTM for text recognition?
Belhakimi, A. M., Chiron, G., Arrestier, F., & Awal, A. M. (2023) · International Conference on Machine Learning Technologies (ICMLT) · doi:10.1145/3589883.3589914
“A previous work of Fedor et al [10] presented a model called Rosetta that uses a very deep residual neural network as a backbone and a CTC.”
MethodDeep Learning-based End-to-End Address Recognition Solution on Chinese Courier Order Forms
Zhang, J., & Liu, Y. (2023) · International Conference on Artificial Intelligence, Automation and Algorithms (AI2A) · doi:10.1145/3611450.3611473
Improved Lightweight Multi-Target Recognition Model for Live Streaming Scenes
Li, Z., Qiao, K., Chen, J., Li, Z., & Zhang, Y. (2023) · Applied Sciences · doi:10.3390/app131810170
Natural Scene Text Detection in Video with Hybrid Text Augmentation and Fusion-Transferred Learning
Devi, M. M., Maddala, S., & Raju, S. V. (2023) · Advances in intelligent systems and computing · doi:10.1007/978-981-99-1588-0_17
A System of Layout Robust Text Detection and Recognition for Washing Label
Qian, L., Xu, S., Yan, M., & Zheng, H. (2023) · IEEE International Conference on Electronic Information and Communication Technology (ICEICT) · doi:10.1109/iceict57916.2023.10245942
Orthogonality-constrained multihead self-attention for scene text recognition
Xu, S., & Zhu, Z. (2023) · Journal of Image and Graphics · doi:10.11834/jig.221049
Shao, L., Zhang, H., Zhang, C., Yan, H., & Sun, Y. (2023) · IEEE International Symposium on Product Compliance Engineering - Asia (ISPCE-ASIA) · doi:10.1109/ispce-asia60405.2023.10365849
“Additionally, Borisyuk et al. [7] introduced curriculum learning on top of CTC, eliminating the need for a separate CHAR model.”
Study on handwritten invoice recognition system
Han, Y.-S., Seo, H.-I., & Seo, D.-H. (2023) · Journal of Advanced Marine Engineering and Technology · doi:10.5916/jamet.2023.47.6.411
Open-Set Text Recognition: Case-Studies
Yin, X.-C., Yang, C. H., & Liu, C. (2024) · SpringerBriefs in computer science · doi:10.1007/978-981-97-0361-6_7
Yin, X.-C., Yang, C. H., & Liu, C. (2024) · SpringerBriefs in computer science · doi:10.1007/978-981-97-0361-6_2
Yin, X.-C., Yang, C. H., & Liu, C. (2024) · SpringerBriefs in computer science · doi:10.1007/978-981-97-0361-6_1
Open-Set Text Recognition Implementations(II): Sample-to-Representation Mapping
Yin, X.-C., Yang, C. H., & Liu, C. (2024) · SpringerBriefs in computer science · doi:10.1007/978-981-97-0361-6_5
Adversarial Sample Generation Method for Scene Text Images Based on Up-samplingPreprint
Zeng, Z., Wang, C., Yang, X., & Chen, Y. (2024) · Research Square · doi:10.21203/rs.3.rs-5287761/v1
Senior Solutions Architect, D. S. D., & Akheel, S. A. (2023) · Progress in Medical Sciences · doi:10.47363/pms/2023(7)e125
Improving Scene Text Recognition with Counting-Aware Contrastive Learning and Attention Alignment
Yang, J., Zhou, B., & Zhu, A. (2024) · Lecture notes in computer science · doi:10.1007/978-981-97-8511-7_10
Deep Learning Based Airport Passenger Baggage Tag Information OCR
Jie, G., Dong, L., & Zhang, P. (2024) · International Conference on Control and Intelligent Robotics (ICCIR) · doi:10.1145/3687488.3687558
Learning to Ask About Text Content in an Image with Fine-Grained Features
Cheng, M., Qiao, W., Chen, J., & Li, X. (2024) · Lecture notes in computer science · doi:10.1007/978-981-96-0567-5_16
Feature enhanced attention decoder for scene text recognition
Xu, M., Zhang, J., Xu, L., Silamu, W., & Li, Y. (2025) · Multimedia Tools and Applications · doi:10.1007/s11042-025-20829-8
DNP: Deep Neural Predictor Attack for Black-Box Scene Text Recognition
Xu, Y., & Dai, P. (2025) · Communications in computer and information science · doi:10.1007/978-981-96-6975-2_28
SVTR-V3: An Improved Model Based on a Single Visual Recognition Network
Lu, Y., & Wang, J. (2025) · Lecture notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering · doi:10.1007/978-3-031-97352-9_26
Context-Aware Multi-Anchor Captioning for Text-Rich Image UnderstandingPreprint
Marchand, T., Roux, L., Kareem, S., & Gauthier, N. (2025) · Preprints.org · doi:10.20944/preprints202510.0736.v1
Context-Guided Multi-Branch Fusion for Text-Dependent Visual Question ReasoningPreprint
Ridder, S., Verbeeck, N., Hensley, C., & Vandenberghe, L. (2025) · Preprints.org · doi:10.20944/preprints202510.1994.v1
Fine-tuning a transformers-based model to extract relevant fields from invoices
Cruz, R. F. P. M. L. D. (2021) · Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT)
Knowledge Distillation Methods for Sequence-to-Sequence Learning in Speech and Language Processing
윤지원 (2024) · Seoul National University Open Repository (Seoul National University)
Extraction et compréhension de texte dans les films
Testeau, S. (2022) · Archipel (Université du Québec à Montréal)
TopoReformer: Mitigating Adversarial Attacks Using Topological Purification in OCR ModelsPreprint
Kumar, B., Aravinthakashan, A. S., Satyanarayan, A., Gakhar, I., & Verma, U. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2511.15807
“The three CTC-based models CRNN (Shi, Bai, and Yao 2016), Rosetta (Borisyuk, Gordo, and Sivakumar 2018), and STAR-Net (Liu et al. 2016) combine convolutional feature extraction with sequence modeling and CTC-based decoding, differing in backbone complexity and feature aggregation.”
BARON ROMERO, M. (2025) · Electronic Theses and Dissertations Repository (University of Pisa)
Spatial pyramid pooling enhanced multi-modal linear transformer for table recognition
Li, H., Qiu, X., Zhang, J., Zhang, B., Lu, B., Wang, C., Zhan, C., & Baishun, S. U. (2025) · Discover Artificial Intelligence · doi:10.1007/s44163-025-00756-1
“With the advancement of deep learning, OCR models based on convolutional neural networks (CNNs) and recurrent neural networks (RNNs), such as CRNN [53] and Rosetta [4], have significantly enhanced recognition robustness through sequence-to-sequence modeling.”
Reimagining CRNN with Attention for Handwritten Chinese Text Recognition in Noisy Backgrounds
Shen, L., Lin, B., Lu, W., Tang, S.-K., & Mirri, S. (2025) · IEEE Symposium on Computers and Communications (ISCC) · doi:10.1109/iscc65549.2025.11325768
“Although deep learning-based text recognition methods have made substantial progress in recent years, most existing work focuses on handwritten or printed text under clean and controlled conditions [4]–[11].”
Ultra-Low-Cost Hybrid OCR–LLM Architecture for Production Grade E-KTP Extraction
Saputro, A. T., Herlambang, B. A., & Novita, M. (2025) · Scientific Journal of Informatics · doi:10.15294/sji.v12i4.38200
“Hybrid OCR–LLM pipelines have been investigated in the wider literature on document intelligence, as an attempt to find a compromise between efficiency and semantic robustness [15], [16].”
AI-Driven Document Intelligence: A Survey on Preprocessing, Detection and Recognition
Cheng, T., Feng, S., Yu, Z., Liu, F., Zhang, P., & Song, P. (2025) · International Conference on Computer, Big Data and Artificial Intelligence (ICCBD+AI) · doi:10.1109/iccbdai66607.2025.11387963
“Subsequently, models such as Rosetta [52] have further optimized the CRNN architecture, adopting designs such as fully convolutional networks to improve efficiency while maintaining performance.”
Metadata
- Authors
- Borisyuk, F., Gordo, A., & Sivakumar, V.
- Affiliations
- Meta (United States) (US)
- Venue
- Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining
- Publisher
- ACM
- Published
- 2018-07-19
- Pages
- 71-79
- OpenAlex
- W2809273748
- Cited by
- 353
- References
- 28
- Citations by year
- 2026: 13 · 2025: 35 · 2024: 44 · 2023: 65 · 2022: 51 · 2021: 79 · 2020: 38 · 2019: 26 · 2018: 2
- FWCI
- 10.92
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Handwritten Text Recognition Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Vehicle License Plate Recognition
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Keywords
- optical character recognition, Rosetta, Facebook, system architecture, text extraction from images, media understanding
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.