Career Profile

I am a AI Research Engineer at Helsing - Artificial Intelligence To Serve Our Democracies at Helsing. Previous to this I was Postdoctoral Researcher in Computer Vision at Computer Vision Center, Universitat Autonoma de Barcelona, Barcelona, Spain. A enthusiastic programmer and a ML/VISION researcher. I am a researcher in Computer Vision, Pattern Recognition, Machine Learning and related fields. Currently, I completed my PhD., under the supervision of Dr. Josep Lladós and Prof. Dr. Umapada Pal. My main research interest includes computer vision, pattern recognition and image processing. In particular, I work with multi-modal data embedding in common sub space, structural pattern recognition, and apply them for various applications such as natural language processing, sketch based image retrieval, object detection, recognition etc.

Experiences

Postdoctoral Researcher

2020 - 2021
CVC, Universitat Autonoma de Barcelona, Spain

Developed various algorithms for Sketch Based Image Retrieval, Image processing and Pattern Recognition. Few mentionable works are as follows

  • SBIR
  • Object Detection

Research Assistant

2015 - 2016
CVC, Universitat Autonoma de Barcelona, Spain

Developed various algorithms for Document Image Analysis, Image processing and Pattern Recognition. Few mentionable works are as follows

  • Signature Verification
  • Word Spotting

Project Linked Personnel

2013 - 2014
CVPR Unit, Indian Statistical Institute, India

Took part in development of various algorithms for Computer Vision, Image processing and Pattern Recognition. Few mentionable works are as follows

  • Heat Kernel Signature
  • Fractals
  • Multi oriented scene text spotting in multi lingual documents

Project Linked Personnel

2011 - 2013
ECS, Unit Indian Statistical Institute, India

Development of stand-alone compact data logger. Study, Design and Guiding Bachelor’s Intern Student.

  • Real-Time Atmospheric Data logger
  • AVR architecture, NI Multisim.

Research Trainee

2011 - 2011
Variable Energy Cyclotron Center, Atomic Energy of India, India

Study, design and develop instrumentation for measurement of magnetic susceptibilty. Observed Frequency-adapted crossover from para-to-dia magnetization in an Ising-like dipole-dipole a model

  • Finite Element Analysis
  • Genetic Algorithm
  • Material Science

Project Engineer

2008 - 2009
Enterprising Developers, India

PLC Drives and Conveyer Belts Installation, Development Standalone Drivers.

  • Real Time Operating System
  • Digital Signal Processing

Projects

Side projects I was associated with during my PhD.

Consolidated Research Group DAG-CVC (SGR) - AGAUR funded project for Image Processing, Pattern Reconition and Computer Vision related problems
Contextual Recognition of Ancient Documents (Co-Read) - MINECO funded project Contextual Recognition of Ancient Documents (Co-READ) studies information spotting techniques to search words and shapes in ancient document images. In addition, contextual search in documents will be investigated. Context means the structural and syntactic relations between basic components (words and symbols)
CONtextualising the COntents in the Recognition of Document Images from Archives (CONCORDIA) - MINECO funded project CONCORDIA aims to make a qualitative leap forward in the interpretation of historical documents with the incorporation of context and knowledge from the experts through advanced human interfaces

Publications

List of my publications in this section is from different domains. Computer Vision, Pattern Recognition, Material Science, Instrumentation and Control. Please refer to my Google Scholar for an updated publication list.

  • Multi-Modal Reasoning Graph for Scene-Text Based Fine-Grained Image Classification and Retrieval
  • Andres Mafla, Sounak Dey, Ali Furkan Biten, Lluis Gomez, Dimosthenis Karatzas
    Winter Application in Computer Vision(WACV), 2021
  • Fine-grained Image Classification and Retrieval by Combining Visual and Locally Pooled Textual Features
  • Andres Mafla, Sounak Dey, Ali Furkan Biten, Lluis Gomez, Dimosthenis Karatzas
    Winter Application in Computer Vision(WACV), 2020
  • Evaluation of word spotting under improper segmentation scenario
  • Sounak Dey, Anguelos Nicolaou, Josep Lladós, Umapada Pal
    International Journal on Document Analysis and Research (IJDAR), 2019
  • Doodle to search: Practical zero-shot sketch-based image retrieval
  • Sounak Dey, Pau Riba, Anjan Dutta, Josep Llados, Yi-Zhe Song
    Computer Vision and Pattern Recognition (CVPR) (ORAL), 2019
  • Aligning Salient Objects to Queries: A Multi-modal and Multi-object Image Retrieval Framework
  • Sounak Dey, Anjan Dutta, Suman K Ghosh, Ernest Valveny, Josep Lladós, Umapada Pal
    Asian Conference on Computer Vision (ACCV), 2018
  • Non-deterministic Behavior of Ranking-based Metrics when Evaluating Embeddings
  • Anguelos Nicolaou, Sounak Dey, Vincent Christlein, Andreas Maier, Dimosthenis Karatzas
    International Workshop on Reproducible Research in Pattern Recognition (ICPR), 2018
  • Learning Cross-Modal Deep Embeddings for Multi-Object Image Retrieval using Text and Sketch
  • Sounak Dey, Anjan Dutta, Suman K Ghosh, Ernest Valveny, Josep Lladós, Umapada Pal
    International Conference on Pattern Recognition (ICPR), 2018
  • Bringing Back Hieroglyph
  • Sounak Dey, Anjan Dutta, Josep Lladós, Umapada Pal
    14th IAPR International Conference on Document Analysis and Recognition (ICDAR), 2017
  • Shallow neural network model for hand-drawn symbol recognition in multi-writer scenario
  • Sounak Dey, Anjan Dutta, Josep Lladós, Alicia Fornés, Umapada Pal
    14th IAPR International Conference on Document Analysis and Recognition (ICDAR), 2017
  • Handwriting recognition by attribute embedding and recurrent neural networks
  • J Ignacio Toledo, Sounak Dey, Alicia Fornés, Josep Lladós
    14th IAPR International Conference on Document Analysis and Recognition (ICDAR), 2017
  • Improving Information Retrieval in Multiwriter Scenario by Exploiting the Similarity Graph of Document Terms
  • Pau Riba, Anjan Dutta, Sounak Dey, Josep Lladós, Alicia Fornés
    14th IAPR International Conference on Document Analysis and Recognition (ICDAR), 2017
  • Improving Information Retrieval in Multiwriter Scenario by Exploiting the Similarity Graph of Document Terms
  • Pau Riba, Anjan Dutta, Sounak Dey, Josep Lladós, Alicia Fornés
    14th IAPR International Conference on Document Analysis and Recognition (ICDAR), 2017
  • Signet: Convolutional siamese network for writer independent offline signature verification
  • Sounak Dey, Anjan Dutta, J Ignacio Toledo, Suman K Ghosh, Josep Lladós, Umapada Pal
    arXiV, 2017
  • Script independent approach for multi-oriented text detection in scene image
  • Sounak Dey, Palaiahnakote Shivakumara, KS Raghunandan, Umapada Pal, Tong Lu, G Hemantha Kumar, Chee Seng Chan
    Neurocomputing, 2017
  • Local binary pattern for word spotting in handwritten historical document
  • Sounak Dey, Anguelos Nicolaou, Josep Llados, Umapada Pal
    Joint IAPR International Workshops on Statistical Techniques in Pattern Recognition (SPR) and Structural and Syntactic Pattern, Recognition (SSPR), 2017
  • Frequency-adapted crossover from para-to-dia magnetization in an Ising-like dipole-dipole model
  • Nilangshu Das, P Barat, Sounak Dey, T Jayakumar
    Science Letters Journal, 2015
  • Design of miniature coil to generate uniform magnetic field
  • Nilangshu K Das, Parthasarathi Barat, Sounak Dey, Tammana Jayakumar
    Progress In Electromagnetics Research, 2014
  • Innovative instrumentation to measure magnetic susceptibility
  • Nilangshu K Das, Sounak Dey, Sarbajit Pal, Parthasarathi Barat
    IEEE Transactions on Magnetics, 2013

    Skills & Proficiency

    Python

    Matlab

    C, C++

    Shell Script

    Embedded C

    VHDL