Scribd Logo
Upload

Welcome to Scribd!

  • 21 views

    Virtual Dressing Room

    Uploaded by

    rabeeanoor
    This document describes a virtual dressing room project created by three students guided by Dr. Ritesh Patel and mentored by Asst. Prof. Mrugendra Rahevar. The project aims to allow users to virtually try on apparel through e-commerce websites before purchasing. It will use image processing and convolutional neural networks to detect body parts and overlay clothing images onto a target body. The major features, development environment, and future enhancements are outlined. References for similar projects using deep learning approaches are also provided.

    Copyright:

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

    Virtual Dressing Room

    Uploaded by

    rabeeanoor
    21 views13 pages
    This document describes a virtual dressing room project created by three students guided by Dr. Ritesh Patel and mentored by Asst. Prof. Mrugendra Rahevar. The project aims to allow users to virtually try on apparel through e-commerce websites before purchasing. It will use image processing and convolutional neural networks to detect body parts and overlay clothing images onto a target body. The major features, development environment, and future enhancements are outlined. References for similar projects using deep learning approaches are also provided.

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document describes a virtual dressing room project created by three students guided by Dr. Ritesh Patel and mentored by Asst. Prof. Mrugendra Rahevar. The project aims to allow users to virtually try on apparel through e-commerce websites before purchasing. It will use image processing and convolutional neural networks to detect body parts and overlay clothing images onto a target body. The major features, development environment, and future enhancements are outlined. References for similar projects using deep learning approaches are also provided.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    21 views13 pages

    Virtual Dressing Room

    Uploaded by

    rabeeanoor
    This document describes a virtual dressing room project created by three students guided by Dr. Ritesh Patel and mentored by Asst. Prof. Mrugendra Rahevar. The project aims to allow users to virtually try on apparel through e-commerce websites before purchasing. It will use image processing and convolutional neural networks to detect body parts and overlay clothing images onto a target body. The major features, development environment, and future enhancements are outlined. References for similar projects using deep learning approaches are also provided.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    of 13

    Virtual Dressing Room

    Prepared by:
    18cCE024 (Pratik Dhoriyani)
    18CE031 (Pruthil Hirpara)
     18CE037 (Dhruv Kachhadiya)

    Guided By : Dr. Ritesh Patel

    U & P U. PATEL DEPARTMENT OF COMPUTER ENGINEERING


    Mentored By: Asst. Prof. Mrugendra Rahevar

    Definition Of Project:

    • Buying apparel through e-commerce has certain limitation


    of trying apparel before buying it. Provide a virtual room to
    try apparel through e-commerce websites before buying it.
    Project Description:

    • Buying apparel through e-commerce has certain limitation of


    trying apparel before buying it.
    • Provide a virtual room to try apparel through e-commerce website
    before buying it.
    • So that users are motivated to buy online apparel through e-
    commerce website and lesser chances of returning and cancelling
    of apparel because of having a facilities to try apparel through a
    virtual room.
    • we were included many things like goggles , t-shirts, ear rings ,
    tiaras , frocks , etc.…
    Motivation:

    • We got this project definition from smart Gujarat


    hackathon’s problem statement .
    • This type of projects are available but those projects were
    made using Kinect motion sensor and we are going to create
    this project using concept of image processing and
    convolutional neural network.
    • Real implementation of that project is not available but as a
    reference lot many research papers are there through which
    at least we can have idea about flow of execution in our
    project.
    Development Environment:
    • Software Requirement:

    • Cmake (3.12.0)
    • Dlib (19.15.0)
    • Opencv (3.4.2.17)
    • Scipy (1.0.0)
    • Cascade trainer gui (1.8.0)
    • Tkinter canvas (8.6.8)
    • Numpy (1.18.1)
    • Anaconda (4.8.2)
    • Python (3.7.4)
    • Flask Web framework (1.1.1)
    Development Environment:

    • Hardware Requirement:
    • For this project we need high configurations system because
    such tasks in this project requires more CPU computation
    and high graphic support to train haar cascade dataset and
    for more accuracy we need to increase number of EPOCH in
    training of convolutional neural network.
    • Once the dataset and network models are trained then we
    can run python code on normal machine also.
    Major Features:
    • First of all for we need to detect body parts of target human body and we need to
    transfer things according to suitable body parts.

    • So for body part detection we were used haar cascade dataset.


    • So what is haar cascade dataset ?
    • 24 * 24 target window is moving on to the image and it contains features
    like(line features,rectenguler feature,edge feature etc..)
    • value of feature is computed using concept of integral images(it would select
    best value of featuure among this by using adaboost classifier)
    • They are like convolutional kernels. haar features are relavent feature for
    object detection and non-relevant features are discarded by adaboost
    algorithm.
    Major Features:
    • How to train haar cascade dataset ?
    • Opencv provides some pre trained haar cascade and for creating
    your own haar cascade there are three methods are available.
    • File provided by Auckland university to create haar cascade
    using command prompt but it is very tedious process and not
    that much of accurate(training time is more).
    • Method provided by youtuber sendex but actually it does not
    uses windows to create haar cascade it uses aws ($5/month).
    • Amin ahemadi gui creater for creating haar cascade (cascade
    trainer gui).
    Major Features:
    • Process of training haar cascade:
    • Haar feature(white black boxes moving on to the image
    and extract the feature.)
    • Integrall images approach(to make fasten the process
    of computation of data)
    • Adaboost classifier (extract best feature and discarding
    non-relevant feature)
    Major Features:
    • We were used flask micro framework as a interface between backend(python) and front
    end(html).And for gui we used tkinter library which was written in python for design user
    interface.
    • @app.route("/") it means whenever the home page is found involk the function below it under
    that function in the end of that function you are returning a function render_template("html
    file") and whatever html file is given in the argument that html page will be shown in the browser
    for that this html file should be available in template folder

    • @app.route("/success",methods=["POST"]
    • it means whenever the route is scuccess then involk the function below it post shows that we
    are sending data to server
    Future Enhancement:
    • After detecting body parts of target image there are
    certain operation we need to perform for make this
    accurate.
    • So for this we need to train three type of network
    through which we can set clothes on target image.
    • PAN (pose alignment network)
    • TRN (texture refinement network)
    • FTN (fitting network)
    Graphical Representation:
    References:
    • M2E-Try On Net: Fashion from Model to Everyone(
    https://arxiv.org/pdf/1811.08599v1.pdf)

    • VITON: An Image-based Virtual Try-on Network(


    https://arxiv.org/pdf/1711.08447.pdf)

    You might also like