newsmode
search
Меню
arrow_back Назад
rss_feedEugene Yan ·27.11.2016 open_in_new

Image classification API is now live!

#Ai Products #Computer Vision #Production Ml
auto_awesomeКраткое саммари

Юджин Ян (Ziyou Yan) объявляет о запуске API классификации изображений на Datagene.io — после нескольких месяцев работы. В отличие от API классификации товаров, работающего с названиями, этот API работает с изображениями товаров, но только в категории моды. Модель построена на Keras и Theano, работает на маленьком сервере AWS без GPU, выдаёт результат менее чем за 300 миллисекунд. Точность валидации: Top-1 — 0,76, Top-5 — 0,974. Позже автор отмечает, что API был отключён ради экономии на облачных расходах. Пост входит в серию материалов о создании API классификации товаров и содержит демонстрацию с примером фото футболки.

translate EN + RU EN RU

eugeneyan

  • Start Here
  • Writing
  • Speaking
  • Prototyping
  • About

    • Image classification API is now live!

    [ deeplearning python production 🛠 ] · 2 min read

    After toiling for a few months on this, product image classification is now live on Datagene.io! While the product classification API works with product titles, the image classification API works with product images, though only for fashion. (Github repositiory)

    Update: API discontinued to save on cloud cost.

    This is part of a series of posts on building a product classification API:

  • Data acquisition and formatting (part 1)
  • Data cleaning and preparation (part 2)
  • API development (part 3)
  • Image classification demo
  • Image search demo

    • Some facts about the image classification API:

  • Works best with e-commerce like fashion images (as that’s what it was trained on)
  • Top-1 validation accuracy: 0.76; Top-5 validation accuracy: 0.974
  • Returns results under 300 milliseconds (will be faster in batch mode with GPU)
  • Built on Keras and Theano, and runs on a tiny AWS server without GPU.

    • How to play with it? First, click on browse to select a file for upload. The file needs to have either a “.png”, “.jpg”, or “.jpeg” extension (case-insensitive).

    A very simple input form for your images, with basic validation.

    Then, click on the submit button below and wait patiently—results should appear in a second or so, depending on network speed and image size.

    A very simple submit button.

    Here’s an example

    Given a photo of a t-shirt…

    I wonder if the model will also have learnt about the tagline on the t-shirt.

    … here’s the result we get. Image classification input t-shirt result

    Oh, they don't have a category for smartass t-shirt.

    It works on mobile too

    Look ma, web responsive!

    You can also try the image classification API on mobile. On iOS, it allows you to take a photo, or select one from your photo library or iCloud drive.

    Please leave suggestions on UI (or any other) improvements in the comments!

    If you found this useful, please cite this write-up as:

    Yan, Ziyou. (Nov 2016). Image classification API is now live!. eugeneyan.com. https://eugeneyan.com/writing/image-categorization-is-now-live/.

    or

    @article{yan2016categorization, title = {Image classification API is now live!}, author = {Yan, Ziyou}, journal = {eugeneyan.com}, year = {2016}, month = {Nov}, url = {https://eugeneyan.com/writing/image-categorization-is-now-live/} }

    Join 11,800+ readers getting updates on machine learning, RecSys, LLMs, and engineering.