`agents.components.vision`¶

Module Contents¶

Classes¶

Vision

This component performs object detection and tracking on input images and outputs a list of detected objects, along with their bounding boxes and confidence scores.

API¶

class agents.components.vision.Vision(*, inputs: List[Union[agents.ros.Topic, agents.ros.FixedInput]], outputs: List[agents.ros.Topic], model_client: Optional[agents.clients.model_base.ModelClient] = None, config: Optional[agents.config.VisionConfig] = None, trigger: Union[agents.ros.Topic, List[agents.ros.Topic], float] = 1.0, component_name: str, **kwargs)¶

Bases: agents.components.model_component.ModelComponent

This component performs object detection and tracking on input images and outputs a list of detected objects, along with their bounding boxes and confidence scores.

Parameters:

inputs (list[Union[Topic, FixedInput]]) – The input topics for the object detection. This should be a list of Topic objects or FixedInput objects, limited to Image (or RGBD) type.
outputs (list[Topic]) – The output topics for the object detection. This should be a list of Topic objects, Detection and Tracking types are handled automatically.
model_client (Optional[ModelClient]) – Optional model client for the vision component to access remote vision models. If not provided, enable_local_classifier should be set to True in VisionConfig This should be an instance of ModelClient. Defaults to None.
config (VisionConfig) – The configuration for the vision component. This should be an instance of VisionConfig. If not provided, defaults to VisionConfig().
trigger (Union[Topic, list[Topic], float]) – The trigger value or topic for the vision component. This can be a single Topic object, a list of Topic objects, or a float value for timed components.
component_name (str) – The name of the vision component. This should be a string and defaults to “vision_component”.

Example usage:

image_topic = Topic(name="image", msg_type="Image")
detections_topic = Topic(name="detections", msg_type="Detections")
config = VisionConfig()
model_client = ModelClient(model=DetectionModel(name='yolov5'))
vision_component = Vision(
    inputs=[image_topic],
    outputs=[detections_topic],
    model_client=model_client
    config=config,
    component_name = "vision_component"
)

take_picture(topic_name: str, save_path: str = '~/emos/pictures', timeout: float = 0.5) → bool¶

Take a picture from a specific input topic and save it to the specified location.

This method acts as an Action to capture a specific frame from a specific camera/topic. It prioritizes triggers over standard inputs if a name conflict exists (though unique names are expected).

Parameters:

topic_name (str) – The name of the topic to capture the image from. Must be one of the component’s registered input topics.
save_path (str) – The directory path where images will be saved. Defaults to “~/emos/pictures”.
timeout (float) – Timeout if an image is not available on the topic. Defaults to 0.5 seconds.

Returns:

True if successful, False otherwise.

Return type:

bool

Raises:

ValueError – If the provided topic_name is not found in inputs.

record_video(topic_name: str, duration: float = 5.0, save_path: str = '~/emos/videos', fps: int = 30) → bool¶

Record a video from a specific input topic for a set duration.

This action spawns a background thread to capture frames and save them to a video file. It does not block the main execution loop.

Parameters:

topic_name (str) – The name of the topic to record from.
duration (float) – The duration of the recording in seconds. Defaults to 5.0.
save_path (str) – The directory path where the video will be saved. Defaults to “~/emos/videos”.
fps (int) – The frames per second for the recording. Defaults to 20.

Returns:

True if the recording thread started successfully, False otherwise.

Return type:

bool

Raises:

ValueError – If the topic_name is not registered.

agents.components.vision¶

Module Contents¶

Classes¶

API¶

`agents.components.vision`¶