Scribd Logo
Upload

Welcome to Scribd!

  • 69 views

    Dunzo Catalog: Team Infinite

    Uploaded by

    sajeet sah
    The document proposes building a catalog database for Dunzo using optical character recognition and natural language processing on receipts and invoices. A custom named entity recognizer would recognize food and grocery items to help with searching and processing. The solution would use React, Node.js, PostgreSQL, Google Cloud Vision API, Python and Spacy and be hosted on GitHub. It is designed to be supervised initially and then made stand-alone once performance is achieved.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Dunzo Catalog: Team Infinite

    Uploaded by

    sajeet sah
    69 views2 pages
    The document proposes building a catalog database for Dunzo using optical character recognition and natural language processing on receipts and invoices. A custom named entity recognizer would recognize food and grocery items to help with searching and processing. The solution would use React, Node.js, PostgreSQL, Google Cloud Vision API, Python and Spacy and be hosted on GitHub. It is designed to be supervised initially and then made stand-alone once performance is achieved.

    Original Title

    Team Infinite

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document proposes building a catalog database for Dunzo using optical character recognition and natural language processing on receipts and invoices. A custom named entity recognizer would recognize food and grocery items to help with searching and processing. The solution would use React, Node.js, PostgreSQL, Google Cloud Vision API, Python and Spacy and be hosted on GitHub. It is designed to be supervised initially and then made stand-alone once performance is achieved.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    69 views2 pages

    Dunzo Catalog: Team Infinite

    Uploaded by

    sajeet sah
    The document proposes building a catalog database for Dunzo using optical character recognition and natural language processing on receipts and invoices. A custom named entity recognizer would recognize food and grocery items to help with searching and processing. The solution would use React, Node.js, PostgreSQL, Google Cloud Vision API, Python and Spacy and be hosted on GitHub. It is designed to be supervised initially and then made stand-alone once performance is achieved.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 2

    Dunzo Catalog

    Team Infinite
    Rajeswari N - Team Leader
    Venkat Raman G

    Solution Summary:

    The solution proposed is to build a catalog database for Dunzo to help the
    organisation build a strong data set that can be used for analytical and marketing
    purposes. The proposed solution involves the following major steps:

    1. Uploading of the receipt/invoice


    2. Use OCR and NLP to process, categorise and get insights from the information
    3. In case of any clarifications, verify from the user to know if the data is extracted
    properly and on any discrepancies, correct the data
    4. Search, visualize and gather insights from the data available

    This solution also proposes a custom built NLP Named Entity Recognizer that is
    capable of recognising various India food items and groceries across all the categories
    which would help make the search and process a lot more efficient. When any new
    items that are not captured by the NER are encountered, the solution is capable of
    adding it to the list of known items and updating the NER which will make the data
    capture of that item in future an easy job.

    Tools Used:

    The solution majorly revolves around full stack development, computer vision
    and natural language processing for the functioning. Below are the languages and
    libraries used to build the solution:

    1. Front end application - React and Material Design


    2. Back end application - Node.js
    3. Database - PostgreSQL
    4. Computer Vision - Google Cloud Vision API
    5. NLP - Python and Spacy

    The solution will be hosted on GitHub and can be cloned and run locally to test
    and deploy the application.
    Technical Architecture:

    Feasibility and Further Steps:

    Accounting into the factors of solution confidence, feasibility and efficiency, this
    solution is designed in such a way that the solution’s processing will be supervised by a
    human in the initial stages to build up the confidence level and once the desired
    performance is achieved, the solution can be made stand-alone.

    This solution also provides a user interface for searching and analytical
    purposes, which can be enhanced further based on need. The architecture follows the
    best principles of microservices, there by making the solution more reliable, reusable
    and scalable.

    For any further queries regarding the solution or the approach, don’t hesitate to reach
    out to one of us at ​gowthamvenkat2605@gmail.com​ or ​rajeswari.it10@gmail.com

    You might also like