In Internal Testing
What it does:
Determines if the camera recorded a clip in color, low light or infrared conditions.
Color and Infrared Video as Conditions for Models
Some models can't run when a camera is recording in infrared mode. For example, it isn't possible to determine the color of a vehicle or a person's clothing because the camera is no longer recording in color. So, this model determines if other recording mode conditional models have the possibility of running.
Transparency about Data Collection:
Whenever we get feedback from our users (or our internal quality control team) on whether a model correctly or incorrectly did what it was supposed to do (e.g. did the person detector detect a person or something else?), we also record the conditions of the lighting during that event. Because low light and infrared views are more difficult for computer vision, we try to make sure our models work in those environments. We use our successes and failures to make a better product. We may develop distinct infrared-only or low light computer vision models in the future, if the daylight trained models perform poorly at night.
This model was made by Survail and will continue to learn.