1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

AI and Fashion Design

Discussion in 'AI - Artificial Intelligence in Digital Marketing' started by Simply Social, Jul 25, 2017.

  1. Simply Social

    Simply Social Jr. VIP Jr. VIP

    Joined:
    Aug 28, 2010
    Messages:
    838
    Likes Received:
    85
    I was reading scientific papers about AI and Fashion Design and read an article about a fashion collection from the IBM Watson Deep Learning Team, that was presented in India. Im intrigued by the idea and since Im in a country, where labour and materials are cheap, i have the idea to integrate AI for designing suits, shirts etc.! Now my question to the experts: Is it a good idea compared to getting a designer, who does it manually? How would the datasets look like? Is it worth the effort? Worth the money investment into a machine learning expert? Im getting designs from a designer pretty cheap, sourced from trends. paying about 10$.
     
    • Thanks Thanks x 1
  2. harold bump

    harold bump Newbie

    Joined:
    Jun 10, 2017
    Messages:
    41
    Likes Received:
    7
    Try amazon machine intelligence with data sets from ebay, you might get someone to scrape the amounts sold, prices, sizes, materials and whatnot and then make predictions using the standard regression model. it'd work even for the marketing info hypothetically to know what sells for the highest price based on titles ect.
     
    • Thanks Thanks x 1
  3. lancis

    lancis Elite Member

    Joined:
    Jul 31, 2010
    Messages:
    1,683
    Likes Received:
    2,426
    Occupation:
    Entrepreneur
    Location:
    Milky Way
    Home Page:
    "AI" in its current state interpolates remarkably well but is unable to extrapolate. Or, at best case has a very limited extrapolation abilities.
    So creativity is not about AI. It can find the best design among the designs you feed in, but it cannot create new designs.
     
    • Thanks Thanks x 1
  4. icarusclickbaron

    icarusclickbaron Newbie

    Joined:
    Jan 13, 2014
    Messages:
    19
    Likes Received:
    4
    If I were you I'd use AWS with the AMI TensorFlow and go to town on image recognition as well as extrapolation -- you'd have to use multiple vectors and plan it out accordingly but Tensorflow is Google and based upon their DeepMind project. PM me with any questions

    If you use the IBM Watson Dev platform be sure to use the image recog system as well as pattern and you'd need to put something together in regard to the manufacturing part. I'd look into Baxter's platform (they have closed betas) as an alternative solution although I know cheap labor exists there for you to use. For IBM Watson / cognitive computing, let me know and ya feel free to PM me!
     
    • Thanks Thanks x 1
  5. HoNeYBiRD

    HoNeYBiRD Jr. VIP Jr. VIP

    Joined:
    May 1, 2009
    Messages:
    7,288
    Likes Received:
    8,260
    Gender:
    Male
    Occupation:
    Geographer, Tourism Manager
    Location:
    Ghosted
    Sounds interesting, but how does this AI designed stuff look like? Is it wearable? I guess OP wants to make ready-to-wear stuff and not runway garments.

    I have a strong feeling that it would be a lot easier and cheaper to just pay for a designer.
     
    • Thanks Thanks x 1
  6. Buzzika

    Buzzika Jr. VIP Jr. VIP

    Joined:
    Jul 8, 2009
    Messages:
    1,161
    Likes Received:
    1,428
    Occupation:
    Hustler
    Location:
    Gurgaon
    Home Page:
    For such a scenario, in order to be successful, you will need loads of data to come to any valuable insights or decisions.
    You cannot get such data without scaling your business, and your business will not scale without such data. Sort of a catch-22 situation if you start with AI from scratch.
     
    • Thanks Thanks x 1
  7. timothywcrane

    timothywcrane Power Member

    Joined:
    Apr 25, 2009
    Messages:
    630
    Likes Received:
    247
    Occupation:
    Internet Promotion Management
    Location:
    USA
    I applaud your interest in AI. The strong suit (pun intended) available to you at the moment would probably be to use ML/AI in your logistics chain. If you are already in the position to acquire inexpensive materials you have a step up, but what data and forecasting about the buyer market? I would first concentrate on getting a structured pattern learning system up and running. Look for insights into structured data by converting it to triples and graphing it, looking for anomolies and trends along the edges, especially newly discovered ones.

    Once you get the hang of this, research into applying ontological reasoning to the data (as the Internet is still yet a semantic search world). Once this is achieved, look into OPPL (ontological pre-processing language) and SWRL. Then you can ramp this data up byapplying Logic programming and reasoning to it. I would recommend using OPPL to cycle SPARQL queries into variabled Epics for your industry domain. It will help keep digital overhead down, but beware, it can also cause overfitting by the reasoner and rabbithole you in the wrong direction and waste your time and resources if not supervised.

    If no overfitting is obvious, then move it through a neuralnetwork and joila... you find out that selling jackets for pet pigmy pigs in Nebraska is the next big fashion trend....

    Most of this is already being done with structured data in MySQL with ML tools like Jena connected. I like shiny things too, but AI might be overkill, especially in the Pre-Training data stage. Simple RDF graphing analysis might get you where you need to be.
     
    • Thanks Thanks x 1
  8. timothywcrane

    timothywcrane Power Member

    Joined:
    Apr 25, 2009
    Messages:
    630
    Likes Received:
    247
    Occupation:
    Internet Promotion Management
    Location:
    USA