Video bounding box annotation is the task of drawing 2D boxes over objects of interest in videos. You provide videos and a list of possible objects to draw bounding boxes over. super.AI identifies any objects you've requested and draws 2D bounding boxes over each of them.
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 demo shows our Video Bounding Box Data Program in action, based on a frequent use case we help our customers solve - car damage estimation in the insurance industry. The video shortly explains how to provide labeling instructions for our Human & AI annotators, as well as showing the 'behind the scenes' labeling inteface that our crowd is using to label your data.
What can I use this for?
Labeled bounding box output data can serve as training data for computer vision algorithms in self-driving cars, drones, and robots, or any other object detection tasks.
How can I try it out?
If you'd like to get a customized video bounding box demo based on your own data points, pleasereach out to our sales teamto get started.