# com.unity.perception The Perception package provides a toolkit for generating large-scale datasets for perception-based machine learning training and validation. It is focused on capturing ground truth for camera-based use cases for now and will ultimately expand to other forms of sensors and machine learning tasks. > com.unity.perception is in active development. Its features and API are subject to significant change as development progresses. [Installation instructions](com.unity.perception/Documentation~/SetupSteps.md) [Setting up your first perception scene](com.unity.perception/Documentation~/GettingStarted.md) ## Package contents |Feature|Description |---|---| |[Labeling](GroundTruth-Labeling.md)|Component which marks a GameObject and its descendants with a set of labels| |[Labeling Configuration](GroundTruth-Labeling.md#LabelingConfiguration)|Asset which defines a taxonomy of labels for ground truth generation| |Perception Camera|Captures RGB images and ground truth from a [Camera](https://docs.unity3d.com/Manual/class-Camera.html)| |[SimulationManager](SimulationManager.md)|Ensures sensors are triggered at proper rates and accepts data for the JSON dataset|