AI Fabric Designer

An innovative, efficient, and time-saving approach using fabric prints data, the system can be defined as "a concept to generate new fabric designs from chosen category of more than 20 available categories that learns fabric designs fed as dataset to the application and generate new ones so that non

Project Title

AI Fabric Designer

Project Area of Specialization

Artificial Intelligence

Project Summary

An innovative, efficient, and time-saving approach using fabric prints data, the system can be defined as "a concept to generate new fabric designs from chosen category of more than 20 available categories that learns fabric designs fed as dataset to the application and generate new ones so that none of the previously fed or generated design is ever re-created, and this could be done within no time”. Every year, industrialists have to put millions of rupees just to pay the designers to come up with something unique where these expensive designers put their months to design while sometimes it comes up with nothing or customers disapproves of that design. AI designer can solve all these problems by creating a new design with it for lesser price, lesser human effort and without taking much time. All a user needs to do is to select a theme or category (for e.g. floral, striped etc.) and the system will train itself from the samples of that category (which is fed to our system as dataset). The outcome is basically a reduced cost and time saving solution to design creation that is a standardized themed design of fabrics, generated after training or from pre-trained model saved in the system. Another value proposition of the system is to compare the newly generated samples with old samples in the dataset and our discriminator model will compare every time to ensure that only new design is generated by the model. This will be done by saving all designs in one place on cloud with a mechanism involved to maintain user’s data privacy. User can also customize the created design by changing it to desired color or alter shape sizes. A complete insight of the two models working in adversary will be maintained by evaluating it based on cost estimation of design generator and discriminator model. This comparison will be helpful in further analysis of how powerful cultural trends adoption is witnessed in different places as well as accuracy of the system

Project Objectives

The objective of the proposed system is to automate textile designers job by an application that offers to implement a radical approach to creation of designs implemented through generative approach of using deep learning in python platform in order to prevent wasting hours on a single design as well as saving money to hire a specific person for designing that understands the context and requirement of your fabric.

Project Implementation Method

wew will be using GANS technology for generating new designs and our product will be SAAS (software as a software) providing login nodes to each user.

Benefits of the Project

Benefits of project include: 

1. time saving of desginers
2. for industrialist they wont need to pay huge salaries to expensive desingers
3. industrials can bring more desgins in market and can beat their coompetitors

Technical Details of Final Deliverable

The methodology used here will be Generative Adversarial Networks (GANs) which is a relatively new Machine Learning architecture for neural networks. You can view GANs and Reinforcement Learning means of improving unsupervised machines (neural networks).  These are models that predict by generating the most likely outcome given a sequence of input samples. As an example, a generative model can generate the next likely design based on the previous fed sample frames. You can view a GAN as a new architecture for an unsupervised neural network able to achieve far better performance compared to traditional nets. To be more precise GANs are a new way of training a neural net. GANs contain not one but two independent nets that work separately and act as adversaries . The first neural net is called the Discriminator (D) and is the net that has to undergo training. D is the classifier that will do the heavy lifting during the normal operation once the training is complete. The second network is called the Generator (G) and is tasked to generate random samples that resemble real samples with a twist rendering them as “new design”. Figure 1 GANs Architecture As an example, consider an image classifier D designed to identify a series of images depicting various logo designs for example. Now consider an adversary (G) with the mission to imitate D using carefully crafted images. This is done by picking a legitimate sample randomly from training set (latent space) and synthesizing a new image by randomly altering its features (by adding random noise). As an example, G can fetch the image of a new design and can add an extra design to the image converting it to a new sample. The result is an image very similar to a normal previously fed image with the exception that it created a new design During training, D is presented with a random mix of legitimate images from training data as well as sample images generated by G. Its task is to identify correct and sample inputs. Based on the outcome, both machines try to fine-tune their parameters and become better in what they do. If D makes the right prediction, G updates its parameters in order to generate better new samples to imitate D. If D’s creation is old, it tries to learn from this mistake to avoid similar repetition in the future. The reward for net D is the number of right new designs and the reward for G is the number D’s errors. This process continues until an equilibrium is established and D’s training is optimized..

Final Deliverable of the Project

Software System

Type of Industry

Others

Technologies

Artificial Intelligence(AI), Cloud Infrastructure

Sustainable Development Goals

Decent Work and Economic Growth, Industry, Innovation and Infrastructure

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com