Marçal Rossinyol (PhD)
CTO & Co-FounderAllRead Machine Learning Technologies
marcal [at] allread [dot] ai
Short Bio
Marçal Rossinyol received his B.Sc. (2004), M.Sc. (2006) and Ph.D. (2009) degrees in Computer Science from the Universitat Autònoma de Barcelona, Spain. From 2009-2020, he was a post-doctoral associate researcher at the Computer Vision Center where he participated in more than 20 competitive research projects, both nationally and from the European Commission, and in more than 15 technology transfer projects, being the IP of several of them.
In 2012 and 2014 he was a Marie Curie fellow researcher at the French company Itesoft and at the Laboratoire Informatique, Image et Interaction of the Université de La Rochelle.
He has served as program committee member in a dozen of international conferences and has been the organizer of several international workshops, competitions and tutorials. He is co-author of more than 95 publications in scientific journals and international conferences. He has won several IAPR best paper awards. With more than 3,000 citations, has an h-index of 32. Google Citations Profile.
He has been a Teaching Assistant and an Adjunct Lecturer at the Computer Science Department of the Universitat Autònoma de Barcelona since 2005. He has received the accreditations for Lecturer and Tenured Assistant from AQU in 2010 and 2017 respectively and the Maître de Conférence accreditation from the Education Ministry from France.
He has participated in several entrepreneurship programs aimed at transfering Intelligent Reading Systems research and innovations from academia to industry. In 2019 he co-founded the spin-off company AllRead where he currently is the CTO. He is a member of the ELLIS Society, the European Laboratory for Learning and Intelligent Systems.
Journal Papers
- VLCDoC: Vision-Language contrastive pre-training model for cross-Modal document classification
- Pay Attention to What You Read: Non-recurrent Handwritten Text-Line Recognition
- Content and Style Aware Generation of Text-line Images for Handwriting Recognition
- Multimodal Grid Features and Cell Pointers for Scene Text Visual Question Answering
- Candidate Fusion: Integrating Language Modelling Into a Sequence-to-Sequence Handwritten Word Recognition Architecture
- EAML: Ensemble Self-Attention-Based Mutual Learning Network for Document Image Classification
- Real-time Lexicon-free Scene Text Retrieval
- On Avoiding Segmentation in Handwritten Keyword Spotting: Overview and Perspectives
- Classificació semàntica i visual de documents digitals
- Feature Extraction by Using Dual-Generalized Discriminative Common Vectors
- Avances en clasificación de imágenes en los últimos diez años. Perspectivas y limitaciones en el ámbito de archivos fotográficos históricos
- Augmented Songbook: an Augmented Reality Educational Application for Raising Music Awareness
- Fast Kernel Generalized Discriminative Common Vectors for Feature Extraction
- La Visió per Computador com a Eina per a la Interpretació Automàtica de Fonts Documentals
- A Study of Bag-of-Visual-Words Representations for Handwritten Keyword Spotting
- Efficient segmentation-free keyword spotting in historical document collections
- A Two-stage Approach to Segmentation-Free Query-by-example Word Spotting
- Multimodal page classification in administrative document image streams
- Flowchart Recognition for Non-Textual Information Retrieval in Patent Search
- Boosting the Handwritten Word Spotting Experience by Including the User in the Loop
- On the Influence of Word Representations for Handwritten Word Spotting in Historical Documents
- Automatic Index Generation of Digitized Map Series by Coordinate Extraction and Interpretation
- Relational Indexing of Vectorial Primitives for Symbol Spotting in Line-Drawing Images
- Symbol Spotting in Vectorized Technical Drawings Through a Lookup Table of Region Strings
- A Performance Evaluation Protocol for Symbol Spotting Systems in Terms of Recognition and Location Indices
Selected Publications in International Conferences and Workshops
- Accelerating Transformer-Based Scene Text Detection and Recognition via Token Pruning
- VLCDoC: Vision-Language contrastive pre-training model for cross-Modal document classification
- Read While you Drive - Multilingual Text Tracking on the Road
- Distilling Content from Style for Handwritten Word Recognition.
- GANwriting: Content-Conditioned Generation of Styled Handwritten Word Images.
- Cross-Modal Deep Networks For Document Image Classification.
- Visual and Textual Deep Feature Fusion for Document Image Classification.
- RoadText-1K: Text Detection and Recognition Dataset for Driving Videos.
- Unsupervised Writer Adaptation for Synthetic-to-Real Handwritten Word Recognition.
- Automatic Structured Text Reading for License Plates and Utility Meters.
- ICDAR 2019 Competition on Scene Text Visual Question Answering.
- Scene Text Visual Question Answering.
- ICDAR 2019 Competition on Scene Text Visual Question Answering.
- Selective Text Style Transfer.
- Self-Supervised Visual Representations for Cross-Modal Retrieval.
- Good News, Everyone! Context driven entity-aware captioning for news images.
- Subtitulació automàtica d'imatges. Estat de l'art i limitacions en el context arxivístic.
- Convolve, Attend and Spell: An Attention-based Sequence-to-Sequence Model for Handwritten Word Recognition.
- Single Shot Scene Text Retrieval.
- Cutting Sayre's Knot: Reading Scene Text without Segmentation. Application to Utility Meters.
- The Robust Reading Competition Annotation and Evaluation Platform.
- Manuscript text line detection and segmentation using second-order derivatives analysis.
- Synthetically generated semantic codebook for Bag-of-Visual-Words based word spotting.
- Field Extraction by hybrid incremental and a-priori structural templates.
- The Robust Reading Competition Annotation and Evaluation Platform.
- SmartDoc 2017 Video Capture: Mobile Document Acquisition in Video Mode.
- LSDE: Levenshtein Space Deep Embedding for Query-by-string Word Spotting.
- Benchmarking Keypoint Filtering Approaches for Document Image Matching.
- Automatic Static/Variable Content Separation in Administrative Document Images.
- Self-supervised learning of visual features through embedding images into text topic spaces.
- Dynamic Lexicon Generation for Natural Scene Images.
- Filtrage de descripteurs locaux pour l'amélioration de la détection de documents.
- Human-Document Interaction - a new frontier for document image analysis.
- Delaunay triangulation-based features for Camera-based document image retrieval system.
- Automatic Verification of Properly Signed Multi-page Document Images.
- Improving Document Matching Performance by Local Descriptor Filtering.
- ICDAR2015 Competition on Smartphone Document Capture and OCR (SmartDoc).
- Towards Query-by-Speech Handwritten Keyword Spotting.
- Novel Line Verification for Multiple Instance Focused Retrieval in Document Collections.
- A Comparative Study of Local Detectors and Descriptors for Mobile Document Classification.
- A Semi-Automatic Groundtruthing Tool for Mobile-Captured Document Segmentation.
- Embedding Document Structure to Bag-of-Words through Pair-wise Stable Key-regions.
- Combining Focus Measure Operators to Predict OCR Accuracy in Mobile-Captured Document Images.
- Normalisation et validation d'images de documents capturées en mobilité.
- Fast structural matching for document image retrieval through spatial database.
- Bag-of-Features HMMs for segmentation-free word spotting in handwritten documents.
- Key-region Detection for Document Images -- Application to Administrative Document Retrieval.
- Document Classification and Page Stream Segmentation for Digital Mailroom Applications.
- Field Extraction from Administrative Documents by Incremental Structural Templates.
- Integrating Visual and Textual Cues for Query-by-String Word Spotting.
- Spotting Graphical Symbols in Camera-Acquired Documents in Real Time.
- Classification of Administrative Document Images by Logo Identification.
- An Interactive Appearance-based Document Retrieval System for Historical Newspapers.
- CVC-UAB's participation in the Flowchart Recognition Task of CLEF-IP 2012.
- Multipage Document Retrieval by Textual and Visual Representations.
- The Role of the Users in Handwritten Word Spotting Applications: Query Fusion and Relevance Feedback.
- Browsing Heterogeneous Document Collections by a Segmentation-free Word Spotting Method.
- Interactive Trademark Image Retrieval by Fusing Semantic and Visual Content.
- Automatic Index Generation of Digitized Map Series by Coordinate Extraction and Interpretation.
- Perceptual Image Retrieval by Adding Color Information to the Shape Context Descriptor.
- Efficient Logo Retrieval Through Hashing Shape Context Descriptors.
- A Kernel-based Approach to Document Retrieval.
- Symbol Recognition by Using a Concept Lattice of Graphical Patterns.
- Logo Spotting by a Bag-of-words Approach for Document Categorization.
- Segmentation Robust to the Vignette Effect for Machine Vision Systems.
- Word and Symbol Spotting Using Spatial Organization of Local Descriptors.
- A Region-Based Hashing Approach for Symbol Spotting in Technical Documents.
- Camera-Based Graphical Symbol Detection.
- Boundary Shape Recognition Using Accumulated Length and Angle Information.
- Symbol Spotting in Technical Drawings Using Vectorial Signatures.
Books and Theses
- Flowchart Recognition in Patent Information Retrieval.
- Graphics Recognition Techniques.
- Interactive Document Retrieval and Classification.
- Symbol Spotting in Digital Libraries: Focused Retrieval over Graphic-rich Document Collections.
- Achievements and New Opportunities in Computer Vision.
- Geometric and Structural-based Symbol Spotting. Application to Focused Retrieval in Graphic Document Collections.
- A Model of Vectorial Signatures in terms of Expressive Sub-shapes: Symbol Indexation in Technical Documents.
Distinctions
- AllRead selected as Top 10 most disruptives Catalan companies in 2022: Exponential Leaders, Catalan Goverment, ministry of business and labour, and ACCIÓ, 2022.
- Membership: ELLIS - the European Laboratory for Learning and Intelligent Systems - Society, 2022.
- AllRead winner of the Emprendedor XXI program under category DeepTechXXI: CaixaBank DayOne, 2021.
- AllRead winner of the Lanzate program: EOI Escuela de Organización Industrial and Orange, 2020.
- Best Paper Award: International Conference on Frontiers in Handwriting Recognition (ICFHR), International Association of Pattern Recognition (IAPR), 2020.
- AllRead winner of the SeedRocket program: Best Spanish startup, BStartup, Banc Sabadell, 2021.
- AllRead selected Project: The Collider, Mobile World Capital Barcelona, 2019.
- Best Poster Award: International Conference on Document Analysis and Recognition (ICDAR), International Association of Pattern Recognition (IAPR), 2017.
- Accreditation: Tenured Assistant, Agència per a la Qualitat del Sistema Universitari de Catalunya (AQU), 2017.
- Winner: Ideas Generation Program, Parc de Recerca de la UAB, 2016.
- Selected Project: Market Assessment Program, ACCIÓ / EADA Business School, 2016.
- Winner: ICFHR Competition on Word Spotting, International Association of Pattern Recognition (IAPR), 2016.
- 2nd place: ICDAR Competition on Word Spotting, International Association of Pattern Recognition (IAPR), 2015.
- Finalist: VALORTEC Competition on Business Ideas, Agència per la Competitivitat de l'Empresa (ACCIÓ), 2014.
- Accreditation: Maître de Confèrences, Ministère de l'Enseignement Supérieur, de la Recherche et de l'Innovation, 2014.
- Finalist: Apps&Cultura Challenge, AppCircus, 2014.
- Winner: Flowchart Recognition Task of the CLEF-IP Evaluation Campaign, CLEF-IP, 2012.
- Best Paper Award: International Conference on Document Analysis and Recognition (ICDAR), International Association of Pattern Recognition (IAPR), 2011.
- Accreditation: Lecturer Teacher, Agència per a la Qualitat del Sistema Universitari de Catalunya (AQU), 2010.