Fig 1.
Raspberry Pi with the core components of WiseEye and examples of sensors and peripherals that could be added.
Fig 2.
(a) Simplified diagram showing inter-connection of components with red lines indicating the power supply and green lines depicting the control/information flow from/to Raspberry Pi. (b) Inside view of WiseEye. The waterproof box has the dimension of 150 x 200 x 80 mm.
Fig 3.
The layout of the roof top study area, showing the positions of the cameras and the detection zone and camera field of view for WiseEye.
The numbers on the grid represent the number and locations of False Positive events outside the field of view of the camera.
Fig 4.
The image processing steps to determine whether an actual target object is present in the field of view by comparing the motion-activated image with a background image.
(a) A time-lapse image (with no target) used for subsequent operations as a background image. The Region of Interest (RoI) to be used for image processing is indicated by a red dashed rectangle. (b) motion-activated image showing the objects of interest and RoI. (c) Difference image between the RoI of (a) and (b), showing the birds as clusters of white “difference” pixels.
Table 1.
Comparative data for WiseEye and Bushnell camera during the three-day roof top trial.
The tabulated numbers refer to the number of images recorded or, in the case of false negatives, the number of potential detections missed.