Introducing VED-X: A Versatile Event Detection Framework

Empower your products with computer vision models that can learn to recognize any visual event that matters from only a few samples.

Cameras

Y Combinator
AWS
NVIDIA

A Novel Few Shot Learning Framework

As humans, we can learn to recognize a new visual event after seeing one or two examples.

We have built a novel Few Shot Learning framework, named VED-X, which is capable of doing just that.

VED-X allows you to train models that can recognize almost any event that can be recognized visually by providing only a few examples of that event.

If the model makes a mistake, you can flag the mistake and your feedback gets incorporated back into the model instantly to improve the accuracy.

"It's like magic!"
That is the most common reaction we get from those who have tried it.

Objects

Demo

In this example, we train a model to detect when a door is open.
In this example, we train a model to detect when a hot tub has been left uncovered, in other words, when the hot tub cover is off and also no one is in/around the hot tub.
In this example, we train a model to detect when a faucet is left running and no one is next to it.
In this example, we train a model to detect whether or not a person is wearing a mask. We also demo how a user can provide feedback to correct the model if it makes a false prediction.
In this example, we train a model to detect when a gas stove is left on.
In this example, we train a model to detect when a package is being picked up.