Thank you for your interest

The super.AI Platform can do almost anything

Want a free trial? You've been selected to use super.AI free for 14 days, we just need a few more details.
Request Your Trial

Instance Segmentation Demo

Want to see your specific use case? 

Schedule a personalized demo

What is Instance Segmentation?

Instance Segmentation is the practice of taking images and identifying, at the level of individual pixels, distinct "things" or foreground objects such as cats, people, cars or trees. It is frequently used in computer vision machine learning projects.

Each instance of, for example, a cat is identified using a binary mask of each pixel in the image (1=belongs to Cat A, 0=does not belong to Cat A). So, if your images shows four cats, you end up with four distinct instances of the class "cat", each with its own binary mask.Instance segmentation is indispensable for highlighting important elements for computer vision tasks, particularly in the field of self-driving cars, but also across drones, robotics and health care.super.AI has the power to turn your raw imagery into labeled data through instance segmentation.

In the example below, each pedestrian is a separate instance of the `pedestrian` class, and each vehicle and lamppost, etc., is likewise a single instance of its own class.

Input image
Output image

What is a data program?

From a high-level perspective, a Data Program is a mechanism to transform any input (e.g. images, video, audio, etc.) to any output (e.g. classifications, categories, etc.). In order to achieve that it can route the data to different labeling sources. Labeling sources can be humans or machines. Parameters, like quality, cost and latency, define to which sources the data is routed.

What does the demo showcase?

The video demo shows our Instance Segmentation Data Program in a frequently used case: estimating car damage in the insurance industry.

What can I use this for?

Instance segmentation can help achieve object detection tasks in real-world scenarios and differentiate between multiple similar objects in the same image. Semantic segmentation can detect objects within the input image, isolate them from the background and group them based on their class. This means the data program can be used for a wide variety of industries, including retail, insurance, self-driving cars and more.

How can I try it out?

If you'd like to get a customized Instance Segmentation demo based on your own data points, please reach out to our sales team to get started.

Up next

NER Demo

Resources

This is some text inside of a div block.

Demo Videos

Bounding Box Demo
Bounding Box Demo
NER Demo
NER Demo
Video Bounding Box Demo
Video Bounding Box Demo
Estimating Car Damage Demo
Estimating Car Damage Demo
Instance Segmentation Demo
Instance Segmentation Demo