Bounding Box

The coordinates of a rectangle that define an ROI within an image. The coordinates are relative to the size of the image (between 0 and 1) and are given as: bbox_x - the x coordinate of the top-left corner of the rectangle bbox_y - the y coordinate of the top-left corner of the rectangle bbox_w - the width of the box as a proportion of the whole image bbox_h - the height of the box as a proportion of the whole image

Camera

A physical camera trap device deployed at a Station.

Capture

Detector

A machine-learning model that is capable of detecting the animals in raw camera trap images (e.g. MegaDetector).

Distance Metric

A method for quantifying the dissimilarity of two vectors.

Cosine Distance

The cosine of the angle between the vectors, or the dot product of the vectors divided by the product of their lengths.

L2 Distance

Also known as the Euclidean distance, the straight-line distance between two vectors in n-dimensional space.

Distance Threshold

The upper-bound of the chosen distance metric that will include an image in the match stack as a potential match when its distance to the query image is below the threshold.

Embedding

The output of the re-identification model that describes the image as a single vector.

Individual

A single, distinct, self-contained living entity. An individual can be given a name, and has attributes sex and age at a given point in time.

Model

In machine learning, a complex algorithm that transforms data of one kind (e.g. a .png image) into another kind (e.g. a 2,180-point vector) after learning patterns of associations through many examples.

Region

A geographical region that may host one or more Surveys (e.g. “Northest Peru”).

ROI

A “Region of Interest” within an image that is defined by a bounding box and typically contains an animal and minimal background.

Sequence

A set of images taken in quick succession at a specific location. A sequence can also refer to the frames of a video.

Station

A specific location within the geological bounds of a Survey that contains one or more camera traps.

Survey

A camera trap study in a particular geographical Region over a specific period of time that encompasses a set of cameras at specific Station locations.

Viewpoint

The visible side of the animal’s body in an image as seen from the camera.