Image acquisition and tracking

Image acquisition

A key feature of Stytra is the extraction of relevant behavioral features in real time from video inputs. The Camera object provides an interface for grabbing frames and setting parameters for a range of different camera types. Currently supported models include those by XIMEA, AVT, PointGray FLIR, and Mikrotron, as well as webcams supported by OpenCV [Bra00]. Support for other cameras can be added as long as a Python or C API exists. In addition, previously-recorded videos can also be processed, allowing for offline tracking. Frames are acquired from the original source in a process separated from the user interface and stimulus display. This ensures that the acquisition and tracking frame rate are independent of the stimulus display, which, depending on the complexity of the stimulus and output resolution, can be between 30 and 60 Hz.

[Bra00]G. Bradski. The OpenCV Library. Dr. Dobb’s Journal of Software Tools, 2000.

Implementation of image processing pipelines

Image processing and tracking pipelines are defined by subclassing the Pipeline class. The pipelines are defined as trees of nodes, starting from the camera image with each node parametrized using lightparam. The image processing nodes are subclasses of ImageToImageNode whereas the terminal nodes are ImageToDataNode

Attributes of pipelines are:

  • a tree of processing nodes, along
  • (optional) a subclass of the camera window which displays the tracking overlay
  • (optional) an extra plotting window class

the nodes can be set as attributes of the class, with names that are arbitrary except for how they are used by the display and plotting classes (see the stytra.experiments.fish_pipelines for examples)

Processing nodes

There are two types of nodes: ImageToImageNode
and ImageToDataNode

Nodes must have:

  • A name
  • A _process function which contains optional parameters

as keyword arguments, annotated with Params for everything that can be changed from the user interface. The _process function has to output a NodeOutput named tuple (from :module:`stytra.tracking.pipelines.pipeline`) which contains a list of diagnostic messages (can be empty), and either an image if the node is a ImageToImageNode or a NamedTuple if the node is a ImageToDataNode

Optionally, if the processing function is stateful (depends on previous inputs), you can define a reset function which resets the state.