diff --git a/NAMESPACE b/NAMESPACE
index 69eb621..1697b7f 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -12,6 +12,8 @@ S3method(write_to_spi_sheet,default)
 S3method(write_to_spi_sheet,image_data)
 S3method(write_to_spi_sheet,sample_station_info)
 export(bccamtrapp)
+export(collapse_migration_episodes)
+export(collapse_wallow_episodes)
 export(fill_spi_template)
 export(fill_spi_template_ff)
 export(make_deployments)
diff --git a/R/collapse.R b/R/collapse.R
new file mode 100644
index 0000000..999cb3b
--- /dev/null
+++ b/R/collapse.R
@@ -0,0 +1,199 @@
+#' Collapse motion detection events into episodes
+#'
+#' @param df a data frame of migration image data, read in with `read_image_data()`.
+#' @param migration_map A small data.frame defining the upstream
+#' direction(s) of travel for each sample station. Valid upstream values are:
+#'  `"L to R"`, `"R to L"`, `"T to B"`, `"B to T"`.
+#' Columns must include:
+#' - `sample_station_label`
+#' - `upstream`
+#'
+#' @returns a data frame with motion detection events collapsed into 1 row per
+#' episode, and summary information about each episode (e.g. duration, number
+#' of triggers, total count, direction of travel, ...).
+#'
+#' @export
+collapse_migration_episodes <- function(df, migration_map) {
+  df <- prep_for_collapse(df)
+  df_collapsed <- collapse_episodes(df, "migration")
+  make_travel_direction(df_collapsed, migration_map)
+}
+
+#' Collapse Wallow Episodes
+#'
+#' @param df a data frame of wallow image data, read in with `read_image_data()`.
+#'
+#' @returns a data frame with wallow image data collapsed into 1 row per
+#' episode, and summary information about each episode (e.g. duration, number
+#' of triggers, total count, social and wallow behaviour, ...).
+#'
+#' @export
+collapse_wallow_episodes <- function(df) {
+  df <- prep_for_collapse(df)
+  df_collapsed <- collapse_episodes(df, "wallow")
+  make_wallow_behaviour_columns(df_collapsed)
+}
+
+prep_for_collapse <- function(df) {
+  if (length(grep("antler_?class", names(df))) > 1) {
+    cli::cli_abort("More than one column with antler class data")
+  }
+
+  df |>
+    dplyr::filter(
+      .data$trigger_mode == "Motion Detection",
+      .data$species != "",
+      !is.na(.data$species)
+    ) |>
+    dplyr::rename(
+      "antler_class" = matches("antler_?class")
+    ) |>
+    dplyr::mutate(
+      episode_num = gsub("^([0-9]+).*", "\\1", .data$episode)
+    )
+}
+
+collapse_episodes <- function(df, which = c("migration", "wallow")) {
+  which <- match.arg(which)
+
+  collapsed_df <- df |>
+    dplyr::group_by(.data$episode_num, .data$species) |>
+    dplyr::summarise(
+      sample_station_label = dplyr::first(.data$sample_station_label),
+      first_date_time = min(.data$date_time, na.rm = TRUE),
+      last_date_time = max(.data$date_time, na.rm = TRUE),
+      duration_s = as.numeric(difftime(
+        .data$last_date_time,
+        .data$first_date_time,
+        units = "secs"
+      )),
+      n_triggers = dplyr::n(),
+      wallow_behaviour_list = if (which == "wallow") {
+        list(unique(stats::na.omit(unlist(strsplit(
+          .data$elk_behaviour_wallow,
+          ",(\\s*)?"
+        )))))
+      },
+      social_behaviour_list = if (which == "wallow") {
+        list(unique(stats::na.omit(unlist(strsplit(
+          .data$elk_behaviour_social,
+          ",(\\s*)?"
+        )))))
+      },
+      direction_of_travel = if (which == "migration") {
+        dplyr::first(.data$direction_of_travel)
+      },
+      dplyr::across(dplyr::where(is.numeric), \(x) sum(x, na.rm = TRUE)),
+      dplyr::across(
+        c(
+          dplyr::where(is.character),
+          -any_of(c(
+            "sample_station_label",
+            "direction_of_travel",
+            "episode"
+          ))
+        ),
+        combine_strings
+      ),
+      dplyr::across(dplyr::where(is.logical), \(x) any(x, na.rm = TRUE)),
+      total_count = sum(.data$total_count_episode, na.rm = TRUE),
+      antler_class_1 = any(grepl("Class 1", .data$antler_class)),
+      antler_class_2 = any(grepl("Class 2", .data$antler_class)),
+      antler_class_3 = any(grepl("Class 3", .data$antler_class)),
+      antler_class_4 = any(grepl("Class 4", .data$antler_class)),
+      .groups = "drop"
+    ) |>
+    dplyr::arrange(.data$first_date_time, .data$episode_num, .data$species)
+
+  collapsed_df
+}
+
+combine_strings <- function(x) {
+  x <- paste(unique(stats::na.omit(x)), collapse = "; ")
+  if (x == "") {
+    return(NA_character_)
+  }
+  x
+}
+
+make_travel_direction <- function(df_collapsed, migration_map) {
+  if (
+    !all(
+      c("sample_station_label", "upstream") %in% names(migration_map)
+    )
+  ) {
+    cli::cli_abort(
+      "migration_map must have columns: sample_station_label, upstream"
+    )
+  }
+
+  df_collapsed |>
+    dplyr::mutate(direction_of_travel = tolower(.data$direction_of_travel)) |>
+    dplyr::left_join(migration_map, by = "sample_station_label") |>
+    dplyr::mutate(
+      migration_direction = dplyr::case_when(
+        .data$direction_of_travel == "unk" ~ NA_character_,
+        .data$direction_of_travel == "enters from top" &
+          .data$upstream == "T to B" ~ "upstream",
+        .data$direction_of_travel == "enters from bottom" &
+          .data$upstream == "T to B" ~ "downstream",
+        .data$direction_of_travel == "enters from top" &
+          .data$upstream == "B to T" ~ "downstream",
+        .data$direction_of_travel == "enters from bottom" &
+          .data$upstream == "B to T" ~ "upstream",
+        .data$direction_of_travel == "enters from left" &
+          .data$upstream == "L to R" ~ "upstream",
+        .data$direction_of_travel == "enters from right" &
+          .data$upstream == "L to R" ~ "downstream",
+        .data$direction_of_travel == "enters from left" &
+          .data$upstream == "R to L" ~ "downstream",
+        .data$direction_of_travel == "enters from right" &
+          .data$upstream == "R to L" ~ "upstream",
+        .default = NA_character_
+      )
+    ) |>
+    dplyr::select(-"upstream", )
+}
+
+make_wallow_behaviour_columns <- function(df_collapsed) {
+  df_collapsed <- df_collapsed |>
+    dplyr::mutate(
+      wallow_behaviour_list = list_to_named_logical(
+        .data$wallow_behaviour_list,
+        prefix = "wallow_"
+      ),
+      social_behaviour_list = list_to_named_logical(
+        .data$social_behaviour_list,
+        prefix = "social_"
+      )
+    )
+
+  df_collapsed |>
+    dplyr::relocate(
+      "wallow_behaviour_list",
+      "social_behaviour_list",
+      .after = "total_count"
+    ) |>
+    tidyr::unnest_wider(
+      .data$wallow_behaviour_list,
+      names_repair = janitor::make_clean_names,
+      transform = \(x) ifelse(is.na(x), FALSE, x)
+    ) |>
+    tidyr::unnest_wider(
+      .data$social_behaviour_list,
+      names_repair = janitor::make_clean_names,
+      transform = \(x) ifelse(is.na(x), FALSE, x)
+    )
+}
+
+list_to_named_logical <- function(x, prefix) {
+  lapply(
+    x,
+    \(y) {
+      nm <- y
+      y <- rep(TRUE, length(y))
+      names(y) <- if (length(nm) == 0) character(0) else paste0(prefix, nm)
+      y
+    }
+  )
+}
diff --git a/inst/scratch-templates.R b/inst/scratch-templates.R
index 154749f..a305a0e 100644
--- a/inst/scratch-templates.R
+++ b/inst/scratch-templates.R
@@ -1,5 +1,6 @@
 # devtools::install_github("bcgov/bccamtrap")
 library(bccamtrap)
+library(dplyr)
 
 migration_data <- read_image_data(
   "inst/extdata/example-data/Example-input-CSV_migration-cams_Migration_v1.csv"
@@ -16,7 +17,23 @@ migration_data_qa <- qa_image_data(
   exclude_human_use = FALSE
 )
 
-migration_data_qa
+migration_map <- tribble(
+  ~sample_station_label , ~upstream ,
+  "Kay-A"               , "R to L"  ,
+  "Kay-B"               , "R to L"  ,
+  "UpperAdam-A"         , "L to R"  ,
+  "UpperAdam-B"         , "R to L"  ,
+  "SwampLake-A"         , "L to R"  ,
+  "SwampLaKe-B"         , "R to L"  ,
+  "SwampLake-B"         , "T to B"  ,
+  "MerrillLk-A"         , "L to R"  ,
+  "Salmon4-A"           , "R to L"  ,
+  "Salmon-A"            , "R to L"  ,
+  "Salmon-B"            , "L to R"  ,
+  "Memekay-A"           , "L to R"
+)
+
+collapse_migration_episodes(migration_data_qa, migration_map = migration_map)
 
 wallow_elk_data <- read_image_data(
   "inst/extdata/example-data/Example-input-CSV_wallow-cams_Elk_Template_Wallows_v1.csv"
@@ -33,7 +50,8 @@ wallow_elk_data_qa <- qa_image_data(
   exclude_human_use = FALSE
 )
 
-wallow_elk_data_qa
+
+collapse_wallow_episodes(wallow_elk_data)
 
 new_wallow_elk_data <- read_image_data(
   "inst/extdata/example-data/Example-input-CSV_wallow-cams_updated_approach.csv"
@@ -49,4 +67,5 @@ new_wallow_elk_data_qa <- qa_image_data(
   exclude_human_use = FALSE
 )
 
-new_wallow_elk_data_qa
+
+collapse_wallow_episodes(new_wallow_elk_data)
diff --git a/man/collapse_migration_episodes.Rd b/man/collapse_migration_episodes.Rd
new file mode 100644
index 0000000..ffe9465
--- /dev/null
+++ b/man/collapse_migration_episodes.Rd
@@ -0,0 +1,28 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/collapse.R
+\name{collapse_migration_episodes}
+\alias{collapse_migration_episodes}
+\title{Collapse motion detection events into episodes}
+\usage{
+collapse_migration_episodes(df, migration_map)
+}
+\arguments{
+\item{df}{a data frame of migration image data, read in with \code{read_image_data()}.}
+
+\item{migration_map}{A small data.frame defining the upstream
+direction(s) of travel for each sample station. Valid upstream values are:
+\code{"L to R"}, \code{"R to L"}, \code{"T to B"}, \code{"B to T"}.
+Columns must include:
+\itemize{
+\item \code{sample_station_label}
+\item \code{upstream}
+}}
+}
+\value{
+a data frame with motion detection events collapsed into 1 row per
+episode, and summary information about each episode (e.g. duration, number
+of triggers, total count, direction of travel, ...).
+}
+\description{
+Collapse motion detection events into episodes
+}
diff --git a/man/collapse_wallow_episodes.Rd b/man/collapse_wallow_episodes.Rd
new file mode 100644
index 0000000..6b21ed6
--- /dev/null
+++ b/man/collapse_wallow_episodes.Rd
@@ -0,0 +1,19 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/collapse.R
+\name{collapse_wallow_episodes}
+\alias{collapse_wallow_episodes}
+\title{Collapse Wallow Episodes}
+\usage{
+collapse_wallow_episodes(df)
+}
+\arguments{
+\item{df}{a data frame of wallow image data, read in with \code{read_image_data()}.}
+}
+\value{
+a data frame with wallow image data collapsed into 1 row per
+episode, and summary information about each episode (e.g. duration, number
+of triggers, total count, social and wallow behaviour, ...).
+}
+\description{
+Collapse Wallow Episodes
+}