Image Processing: Easily Solving Image Data-Driven Business Problems

Posted by

The new BigML release is here! Join us on Wednesday, December 15 at 8:30 AM PST / 10:30 AM CST / 5:30 PM CET for a FREE live webinar to discover the enhanced version of our Machine Learning platform. Now, BigML Image Processing lets you solve a wide variety of computer vision and image classification and regression problems, including use cases in medical image analysis, visual product search, security surveillance, and vehicle damage detection, among others. All in a single platform with remarkable ease of use!

BigML Image Processing

The BigML Team is thrilled to release Image Processing, as an important milestone towards our mission of making Machine Learning easy and beautiful for everyone. This latest feature will boost your productivity significantly as you look to generate valuable business insights from your image data. Now, you can easily label image data, train and evaluate supervised and unsupervised models using it, make predictions, and automate end-to-end Machine Learning workflows on a single platform

As image capture and storage has become much more affordable, there are many industries and use cases where Image Processing is having a great impact. In healthcare, for instance, BigML analyzes medical images to help doctors accurately diagnose specific illnesses. In retail, image classification is key for e-commerce websites with large image datasets as it speeds up product discoverability. In the security industry, closed-circuit television images help detect intrusions and raise timely alerts. The transportation industry applies Image Processing to control speed limits as well as to avoid accidents. The manufacturing industry utilizes it to automatically detect defective products. The list of examples goes on and can be expanded to cover almost all industries.

How does Image Processing work?

BigML treats images as any other input field. This unique implementation allows you to use image data alongside text, categorical, numeric, date-time, and items data types as input to create any Machine Learning model available in our platform, both supervised and unsupervised, providing ease of deployment, immutability, traceability, and programmability also for images

You will be able to work with your image data from the BigML Dashboard, via the API, or with WhizzML for end-to-end automation. To make this possible, we are bringing composite sources to the platform. A composite source is a collection of component sources supporting multiple formats, including not only tables, but also individual images. You can then create a dataset from any collection of images, and use them for modeling purposes, as any other dataset. This streamlines your image dataset management and allows combining image inputs with other fields, such as labels for classification, numeric features extracted from the image, or other metadata. 

How can you benefit from Image Processing?

When it comes to modeling, BigML offers a variety of ML capabilities on image data, designed to accommodate many different use cases. If your data is suited to deep learning, you can train a convolutional neural network (CNNs) from scratch, or leverage the power of transfer learning by starting from CNN which has been pre-trained on millions of data points and thousands of classes. For cases where off-the-shelf deep learning may not be the best tool, BigML also offers a number of image feature extraction techniques optimized for speed and performance in certain use cases.

As usual, BigML’s goal is to let you focus on what matters the most: your business problem. This means not having to worry about running specialized GPU servers, assuring compatibility between countless software libraries, and micro-managing system resources. BigML eliminates all that complexity by automatically allocating system resources to optimize image processing tasks.

Do you want to know more about Image Processing?

Please visit the dedicated release page and join the FREE live webinar on Wednesday, December 15 at 8:30 AM PST / 10:30 AM CST / 5:30 PM CET. Register today, space is limited!

Finally, stay tuned for the upcoming series of blog posts about Image Processing, where we will start by explaining the basic concepts of this new feature, and continue showcasing it with example use cases, and tutorials on how to use Image Processing through the BigML Dashboard, API, and the Python Bindings, as well as how this new feature has been implemented on the BigML platform.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s