People say that machine learning is going to change the way the internet works. It will make businesses smarter, more effective, and more informed. There will be higher click-through-rates, better content and product recommendation, and better customer segmentation.
The reality in 2017 is that there is a lack of programmers, more complexity, and a lack of quality tools. Generally, machine learning is raw, so no one wants to use it until it doubles revenue.
I come from an ML SaaS (Machine Learning Software as a Service) company. …
All about Core ML models in production
Core ML is new machine learning framework from Apple. It brings machine learning models to Apple devices and makes it easy for developers to take an advantage of ML. We can use a dozen of models prepared by Apple or convert the open-sourced model from popular ML frameworks like Keras, Caffe or scikit-learn. To run an ML powered application in production there is much more to cover. In this post, I want to cover what I learned about ML production flow and challenges of machine learning powered projects.
Machine learning is an iterative process
And core Core ML is a small part of this iterative process…
New homepage, millions of classifications and powerful backend features almost ready for beta access.
August was connected with scaling. We are fascinated with how many calls we received and how AWS helped us to scale. We also finished a new homepage.
Architecture redesign with Kubernetes
As a small startup we started lightweight and now we are getting to the point when scaling is the most painful and joyful task for us. We increased processing power and we can now process 8 images per model in parallel. As long as we run our models in docker containers our next step is to implement advanced container orchestration by Kubernetes. …
Backend updates, increasing number of calls and image upload via URL
It has been a busy month in our Prague office. Two new team members joined us and we are at full throttle to get rid of the BETA label. We experienced a couple of service outages connected to scaling and increasing number of API calls.
Send image as URL
This month we added a possibility to send API calls with an link to the image. You can simply replace a “image” for “url” and send a link as a value. This comes handy for the users who store their images in cloud services like AWS and Google Cloud.
You can also use both cl-api.vize.ai…
Tips for developers building image recognition solution with small dataset
Machine learning is still young but it becomes highly available to developers. Apple Vision SDK, Google Tensorflow or Vize image classifier make it easy to train custom recognition models. While running a vision platform I get many questions and we deal with limitations of technology. This post is for people who are building image classifier to help them define the task to get the most out of today's technology.
I had a conversation with a client who said that google vision does not work and it returns some not relevant…
Human analogy to describe machine learning in image classification
I could point to dozens of articles about machine learning and convolutional neural networks. Every article describes different details. Sometimes too many details are mentioned and so I decided to write my own post using the parallel of machine learning and the human brain. I will not touch any mathematics or deep learning details. The goal is to stay simple and help people experimenting with Vize.ai to meet their goals.
Introduction
Machine learning provides computers with the ability to learn without being explicitly programmed.
For images: We want something that can look at a set of images and remember the patterns…
Recognize images using Vize Classifier API
Today I will show how to set and test custom image classification engine using Vize.ai — Custom image classification API. We will prepare dataset, upload images, train classifier and test our classifier in the web interface. We need no coding experience unless we want to build API in our project. Let’s start now.
Prepare dataset
We want classifier to recognize person’s gender. First, we will need 20 images of each male and female. Let’s google “man” and “woman” and save 20 images of each. Here you can download my dataset. I tried to avoid fashion images and models so my dataset is…
How to plan computer vision features and choose the right provider
Earlier, I wrote a post about the difference between general and custom computer vision platforms. Today I would like to focus on real world use-case. Lets dive into image recognition features planning.
Imaginary use-case
We are running a small business to sell colourful socks. We want to add a “Socks Matching Engine” feature into our app. Customers will upload a picture of two different socks and app tells them how cool the combination is. This is sometimes called a visual search. This way we are going to gain a respect of Hollywood’s fashion policy and increase the sales!
Plan visual image recognition task
Before we start let’s…
New Preview feature in Vize app, no code required.
No more testing using Postman or python script. We want our vision service to be easy to use. With new Vize Preview feature you can test your custom models on different devices. No code required! Simply share and present your model to colleagues and coworkers.
We made it easy to test and share your API. Just click on Preview button after your task is trained and upload image. You can access Preview on both mobile and desktop device. Take a picture on your phone, Drag and drop on desktop.
New opportunities in food industry automation and science
Automotive, electronic manufacturing, mechanical engineering are always searching for a way to automate repeating tasks. In this post, I want to mention three segments where current progress in AI enabled new forms of automation.
Why is machine learning good for industries? Because not every part of the process is precise. Products are not always aligned in a grid and do not have same colour and shape. This is hard to cover by standard visual systems.
Here comes the power of machine learning. We do not need to add more rules to our process line. We only need to gather enough…
