figshare
Browse
ARCHIVE
RS_Interface_Data.zip (1.22 GB)
VIDEO
flock_0.avi (16.83 MB)
IMAGE
flock_0.jpg (518.8 kB)
VIDEO
flock_1.avi (19.08 MB)
IMAGE
flock_1.jpg (864.86 kB)
VIDEO
flock_2.avi (19.96 MB)
IMAGE
flock_2.jpg (640.26 kB)
VIDEO
flock_3.avi (25.9 MB)
IMAGE
flock_3.jpg (863.19 kB)
VIDEO
flock_4.avi (17.52 MB)
IMAGE
flock_4.jpg (712.32 kB)
IMAGE
flock_5-2.jpg (670.71 kB)
VIDEO
flock_5.avi (21.58 MB)
IMAGE
flock_5.jpg (633.09 kB)
VIDEO
flock_6.avi (21.82 MB)
IMAGE
flock_6.jpg (873.11 kB)
1/0
16 files

Supplementary material and code for paper entitled "Simultaneous Measurements of Three-Dimensional Trajectories and Wingbeat Frequencies of Birds in the Field"

Download all (1.36 GB) This item is shared privately
dataset
modified on 2019-03-12, 18:10
Supplementary material and code for “Simultaneous Measurement of Three-dimensional Trajectories and Wingbeat Frequencies of Birds in the Field”

Citation: H. Ling, G. E. McIvor, G. Nagy, S. MohaimenianPour, R. T. Vaughan, A. Thornton, and N. T. Ouellette, "Simultaneous measurements of three-dimensional trajectories and wingbeat frequencies of birds in the field," J. R. Soc. Interface. (https://doi.org/10.1098/rsif.2018.0653).

Contacts:
Nicholas Ouellette, Stanford University, nto@stanford.edu
Hangjian Ling, Stanford University, linghj@stanford.edu

The supplementary material and code contains the following items:

1. Folder ‘code’ includes Matlab code to:
(a) detect birds on images using ‘Step1_BirdDetection.m’;
(b) reconstruct birds’ 3D locations using ‘Step2_Recon3D.m’. The stereo-matching algorithm solves the optical occlusion problem frame by frame.
(c) track individuals’ 3D motion using ‘Step3_tracking.m’. The tracking code uses a three-frame predictive particle tracking algorithm that uses estimates of both velocity and acceleration. Details of the algorithm can be found in Ouellette, et al., (2006). A quantitative study of three-dimensional Lagrangian particle tracking algorithms. Experiments in Fluids 40, 301-313.
(d) calculate wing motion and wingbeat frequency from tracking results using ‘Step4_wing.m’.

Sample raw images and the corresponding camera parameters are provided to test these codes.

2. Folder ‘Data_raw’ includes sample images captured by four cameras. The images from each camera are contained in a separate folder labelled by the camera number, which each contain 110 images (where the background has been removed). The birds are represented by dark pixels on the images.

The camera parameters for each camera is stored in the file ‘camcoefs.mat’. They are represented by 11 coefficients obtained by direct linear transformation. For more details on the camera calibration method, refer to Theriault, et al., (2014). A protocol and calibration method for accurate multi-camera field videography. Journal of Experimental Biology 217(11), 1843–1848

These raw data can be used to test the Matlab codes. There are 6 trajectories in this dataset.

Folder ‘Data_raw’ also includes six videos corresponding to six flock events captured by one of the four cameras. The six flock events correspond to the six flock data presented in the paper.

3. Folder ‘Data_processed’ includes plain text files and video files. The plain text files include 6 flocks consisting of 50 to 300 jackdaws and rooks, 32 jackdaws and 6 rooks flying alone or in isolated pairs. Each text files includes bird id number, positions, times, velocities, accelerations, and wingbeat frequencies at every time step. The video files include six flocks corresponding to the plain text files. These data were used to produce the results in the paper.

Funding

Human Frontier Science Program