This is an example of processing data from a swimming rock prickleback, Xiphister mucosus.

The data was tracked using Sleap (https://sleap.ai/) and comes out in the following format. frame_idx is the frame number, and each point along the body is identified with the point name and .x, .y, the coordinate, and .score, which is a measure of the estimated accuracy of the point. All of the points together are also given a score (instance.score).

head(xmucosusdata)
#> # A tibble: 6 × 27
#>   track frame_idx instance.score Snout.x Snout.y Snout.score BP1.x BP1.y
#>   <lgl>     <dbl>          <dbl>   <dbl>   <dbl>       <dbl> <dbl> <dbl>
#> 1 NA            0           6.08   1753.    821.       0.904 1864.  833.
#> 2 NA            1           6.07   1752.    821.       0.922 1861.  833.
#> 3 NA            2           6.07   1752.    821.       0.919 1860.  833.
#> 4 NA            3           6.06   1749.    821.       0.928 1857.  833.
#> 5 NA            4           6.09   1748.    821.       0.926 1857.  833.
#> 6 NA            5           6.01   1745.    820.       0.903 1856.  833.
#> # ℹ 19 more variables: BP1.score <dbl>, BP2.x <dbl>, BP2.y <dbl>,
#> #   BP2.score <dbl>, BP3.x <dbl>, BP3.y <dbl>, BP3.score <dbl>, BP4.x <dbl>,
#> #   BP4.y <dbl>, BP4.score <dbl>, BP5.x <dbl>, BP5.y <dbl>, BP5.score <dbl>,
#> #   BP6.x <dbl>, BP6.y <dbl>, BP6.score <dbl>, Tail.x <dbl>, Tail.y <dbl>,
#> #   Tail.score <dbl>

Rearrange the data

We need to rearrange the data into “long” form, where the point names are in a separate column, and we have single columns for the x, y, and score values. The function pivot_kinematics_longer is a convenience wrapper for tidyr::pivot_longer that understands the format of points from Sleap or DeepLabCut.

You need to pass in the names of your points (in order from head to tail) and the name of the column you want to put the point names in.

pointnames <- c("Snout", "BP1", "BP2", "BP3", "BP4", "BP5", "BP6", "Tail")
xmucosusdata <- xmucosusdata |> 
  pivot_kinematics_longer(pointnames = pointnames,
                          point_to = "bodypart")

head(xmucosusdata)
#> # A tibble: 6 × 7
#>   track frame_idx instance.score bodypart     x     y score
#>   <lgl>     <dbl>          <dbl> <fct>    <dbl> <dbl> <dbl>
#> 1 NA            0           6.08 Snout    1753.  821. 0.904
#> 2 NA            0           6.08 BP1      1864.  833. 0.797
#> 3 NA            0           6.08 BP2      1928.  828. 0.677
#> 4 NA            0           6.08 BP3      2001.  813. 0.768
#> 5 NA            0           6.08 BP4      2081.  804. 0.603
#> 6 NA            0           6.08 BP5      2157.  816. 0.633

Process the data

From here on out, the steps to process the data are quite similar to what is shown in the main vignette.

xmucosusdata |> 
  filter(frame_idx %in% c(10, 20)) |> 
  mutate(frame_idx = factor(frame_idx)) |> 
  ggplot(aes(x = x, y = y, color = frame_idx, group = frame_idx)) +
  geom_path() +
  coord_fixed()

Get arc length

xmucosusdata <-
  xmucosusdata |> 
  group_by(frame_idx) |> 
  mutate(arclen0 = arclength(x, y))

xmucosusdata |> 
  ungroup() |> 
  filter(bodypart %in% c("BP2", "BP5", "Tail")) |> 
  ggplot(aes(x = frame_idx, y = arclen0, color = bodypart)) +
  geom_point()

Interpolate to constant arc length

xmucosusdata <- xmucosusdata |> 
  interpolate_points_df(arclen0, x, y, spar = 0.2,
                        tailmethod = 'extrapolate',
                        .frame = frame_idx,
                        .point = bodypart,
                        .out = c(arclen='arclen', xs='x_s', ys='y_s'))

xmucosusdata |> 
  filter(frame_idx %in% c(10, 20)) |> 
  mutate(frame_idx = factor(frame_idx)) |> 
  ggplot(aes(x = x, y, color = frame_idx, group = frame_idx)) +
  geom_point(shape = 10) +
  geom_path() +
  geom_point(aes(x = x_s, y = y_s), shape = 5) +
  coord_fixed()

Interpolate the width

xmucosusdata <-
  xmucosusdata |> 
  group_by(frame_idx) |> 
  mutate(width = interpolate_width(fishwidth$s, fishwidth$eelwidth, arclen))

Get the center

xmucosusdata <-
  xmucosusdata |> 
  ungroup() |> 
  get_midline_center_df(arclen, x_s,y_s, width=width,
                        .frame = frame_idx)
#> ℹ Estimating center of mass based on width

Get the swimming axis

xmucosusdata <-
  xmucosusdata |> 
  mutate(x_ctr = x_s - xcom,
         y_ctr = y_s - ycom,
         t = frame_idx / 60) |> 
  get_primary_swimming_axis_df(t, x_ctr,y_ctr,
                               .frame = frame_idx,
                               .point = bodypart)

xmucosusdata |> 
  filter(frame_idx %in% c(10, 20, 30)) |> 
  ggplot(aes(x = x, y = y, color = frame_idx)) +
  geom_path(aes(group = frame_idx)) +
  geom_segment(data = ~ filter(.x, bodypart == "Tail"), 
               aes(x = xcom, y = ycom, 
                   xend = xcom - 50*swimaxis_x, 
                   yend = ycom - 50*swimaxis_y)) +
  geom_point(data = ~ filter(.x, bodypart == "Tail"), 
               aes(x = xcom, y = ycom), color = 'red') +
  facet_grid(frame_idx ~ .) +
  coord_fixed()