Classification Models: How to organize images with computer vision

Now that most smartphones come fully equipped with high-quality cameras—and more than 90 percent of the population owns a mobile device — many of us are carrying around thousands of photo files in our back pockets. While phones do offer some simple tags to help users sort through their personal archives, computer vision classification models offer an easy and effective way of creating custom, highly-specialized models that can help to categorize visual data with highly personalized labels. 

Plainsight’s vision AI On-Demand platform for SaaS makes creating custom classification models fast, easy, and—best of all–you don’t need to be a computer vision expert! The learning curve is shallow for anyone with experience sharing their images online, as Plainsight  offers an intuitive interface that guides users through custom classification without requiring a crash course in coding. 

In this post, we’ll guide you through the process of creating a classification model in 6 easy steps.

What is a Classification Model, and Why Does it Matter?

A classification model is a computer vision algorithm that has been trained to categorize an image. The model learns how to do this by continuously examining images that have been labeled manually until it has processed enough labeled data to accurately categorize new images automatically. This is extremely useful for sorting images into searchable categories. 

How to build a classification model

1. Choose your data: I am an avid hiker and love to visit Yosemite National Park. In addition to towering granite walls, Yosemite has some of the most beautiful waterfalls in the world. I wanted a way to be able to sort my numerous photos of the different waterfalls so that I could easily categorize them. A computer vision classification model is perfect for this! 

2. Uploading your data: With Plainsight’s On-Demand platform I was able to transfer 143 waterfall photos from my local device into my Yosemite Waterfalls dataset project folder.

3. Classification Labels: Next, I created label tags for the waterfalls I wanted to classify. For this project I had classification labels for 8 distinct waterfalls: Yosemite Falls, Upper Yosemite Falls, Lower Yosemite Falls, Vernal Fall, Nevada Fall, Illilouette Falls, and Chilnualna Falls.

4. Labeling the Data: Going through each photo I assigned the correct classification label until the entire dataset was labeled. I am very familiar with what the waterfalls look like so I spent an average of 2 seconds on each image.

5. Training the model: Once the dataset is tagged and submitted, Plainsight’s SmartML feature makes it easy to train an accurate model. After locking my labeled dataset into a version, I hit the ‘train model’ command and waited…

6. The Results: After only 15 minutes and with less than 200 photos I was able to create a classification model for Yosemite Waterfalls with an 83% accuracy. I was pretty impressed! To have such high accuracy on the first version of the model with such a small amount of data means that I’m on the right track to be successful in creating a highly accurate classification model. 

With a working classification model I can now easily sort and categorize any future photos that I take of waterfalls in Yosemite–which can really be a time-saver! 

Computer vision classification models can be used to help image archivists, whether amautuer or professional, automate organization and save time when searching for specific images. 

Plainsight On-Demand users can create unlimited classification labels and easily train custom classification models with our no code SmartML model training feature. Get started by signing up now. For enterprises or large volume archivists, please schedule a demo with our team to discuss your needs and the ways we can help you transform your visual library.

More Plainsight Blog Posts:

Chips and Science Act to Pump $52 Billion into U.S. Manufacturing

Chips and Science Act to Pump $52 Billion into U.S. Manufacturing

Throughout the pandemic, “supply chain woes” have been blamed for everything from empty store shelves and skyrocketing consumer prices to a more recent surplus in inventory at big-box stores. All of this has many experts unsure about how to characterize the current state of the U.S. economy.

What is machine vision?

What is machine vision?

Machine vision was really born on the assembly line, designed as a system of existing technologies and machinery that, combined, “watch” the production process and recognize when flaws occur.