docs: Add comprehensive RF100 dataset catalog with search functionality

DESCRIPTION

@@ -57,7 +57,9 @@ Suggests:
     testthat,
     coro,
     R.matlab,
-    xml2
+    xml2,
+    knitr,
+    rmarkdown
 BugReports: https://github.com/mlverse/torchvision/issues
 Collate: 
     'folder-dataset.R'
@@ -88,8 +90,8 @@ Collate:
     'imagenet.R'
     'models-alexnet.R'
     'models-convnext.R'
-    'models-convnext_segmentation.R'
     'models-convnext_detection.R'
+    'models-convnext_segmentation.R'
     'models-deeplabv3.R'
     'models-efficientnet.R'
     'models-efficientnetv2.R'
@@ -106,6 +108,7 @@ Collate:
     'models-vit.R'
     'ops-box_convert.R'
     'ops-boxes.R'
+    'rf100-catalog.R'
     'tiny-imagenet-dataset.R'
     'transforms-array.R'
     'transforms-defaults.R'
@@ -114,3 +117,7 @@ Collate:
     'transforms-tensor.R'
     'utils.R'
     'vision_utils.R'
+Depends: 
+    R (>= 3.5)
+LazyData: true
+VignetteBuilder: knitr

NAMESPACE

@@ -96,12 +96,14 @@ export(flickr30k_caption_dataset)
 export(flickr8k_caption_dataset)
 export(flowers102_dataset)
 export(generalized_box_iou)
+export(get_rf100_catalog)
 export(image_folder_dataset)
 export(imagenet_classes)
 export(imagenet_label)
 export(kmnist_dataset)
 export(lfw_pairs_dataset)
 export(lfw_people_dataset)
+export(list_rf100_datasets)
 export(magick_loader)
 export(mnist_dataset)
 export(model_alexnet)
@@ -191,6 +193,7 @@ export(rf100_infrared_collection)
 export(rf100_medical_collection)
 export(rf100_peixos_segmentation_dataset)
 export(rf100_underwater_collection)
+export(search_rf100)
 export(tensor_image_browse)
 export(tensor_image_display)
 export(tiny_imagenet_dataset)

NEWS.md

@@ -1,5 +1,9 @@
 # torchvision (development version)
 
+## New features
+
+* Added RF100 dataset catalog with `search_rf100()`, `get_rf100_catalog()`, and `list_rf100_datasets()` functions for discovering and exploring RoboFlow 100 datasets (#271, @ANAMASGARD).
+
 ## New models
 
 * Added `model_convnext_*_detection()` for object detection, with * within tiny/small/base (#262, @ANAMASGARD).

R/collection-rf100-biology.R

@@ -3,7 +3,7 @@ NULL
 
 #' RoboFlow 100  Biology dataset Collection
 #'
-#' Loads one of the [RoboFlow 100 Biology](https://universe.roboflow.com/browse/biology) datasets with COCO-style
+#' Loads one of the [RoboFlow 100 Biology](https://universe.roboflow.com/browse/biology) datasets with
 #' bounding box annotations for object detection tasks.
 #'
 #' @inheritParams rf100_document_collection

R/collection-rf100-damage.R

@@ -3,7 +3,7 @@ NULL
 
 #' RoboFlow 100 Damages dataset Collection
 #'
-#' Loads one of the [RoboFlow 100 Damage & Risk assesment](https://universe.roboflow.com/browse/damage-risk) datasets with COCO-style
+#' Loads one of the [RoboFlow 100 Damage & Risk assesment](https://universe.roboflow.com/browse/damage-risk) datasets with
 #' bounding box annotations for object detection tasks.
 #'
 #' @inheritParams rf100_document_collection

R/collection-rf100-doc.R

@@ -3,7 +3,7 @@ NULL
 
 #' RoboFlow 100 Document dataset Collection
 #'
-#' Loads one of the [RoboFlow 100 Document](https://universe.roboflow.com/browse/documents) datasets with COCO-style
+#' Loads one of the [RoboFlow 100 Document](https://universe.roboflow.com/browse/documents) datasets with
 #' bounding box annotations for object detection tasks.
 #'
 #' @param dataset Dataset to select within \code{c("tweeter_post", "tweeter_profile", "document_part",
@@ -14,7 +14,7 @@ NULL
 #' @param target_transform Optional transform function applied to the target.
 #'
 #' @return A torch dataset. Each element is a named list with:
-#' - `x`: H x W x 3 array representing the image.
+#' - `x`: H x W x 3 array representing the image, auto-oriented and stretched to 640 x 640.
 #' - `y`: a list containing the target with:
 #'     - `image_id`: numeric identifier of the x image.
 #'     - `labels`: numeric identifier of the N bounding-box object class.

R/collection-rf100-infrared.R

@@ -3,8 +3,8 @@ NULL
 
 #' RoboFlow 100 Infrared dataset Collection
 #'
-#' Loads one of the [RoboFlow 100 Infrared](https://universe.roboflow.com/browse/infrared) datasets (COCO
-#' format) with per-dataset folders and train/valid/test splits.
+#' Loads one of the [RoboFlow 100 Infrared](https://universe.roboflow.com/browse/infrared) datasets
+#' with per-dataset folders and train/valid/test splits.
 #'
 #' @inheritParams rf100_document_collection
 #' @param dataset Dataset to select within \code{c("thermal_dog_and_people", "solar_panel", "thermal_cheetah", "ir_object")}.

R/collection-rf100-medical.R

@@ -3,8 +3,8 @@ NULL
 
 #' RoboFlow 100 Medical dataset Collection
 #'
-#' Loads one of the [RoboFlow 100 Medical](https://universe.roboflow.com/browse/medical) datasets (COCO
-#' format) with per-dataset folders and train/valid/test splits.
+#' Loads one of the [RoboFlow 100 Medical](https://universe.roboflow.com/browse/medical) datasets
+#' with per-dataset folders and train/valid/test splits.
 #'
 #' @inheritParams rf100_document_collection
 #' @param dataset Dataset to select within \code{c("radio_signal",

R/globals.R

@@ -1 +1 @@
-utils::globalVariables(c("..", "N", "magick_loader", "self"))
+utils::globalVariables(c("..", "N", "magick_loader", "self", "rf100_catalog"))

R/rf100-catalog.R

@@ -0,0 +1,176 @@
+#' RF100 Dataset Catalog
+#'
+#' A comprehensive catalog of all RF100 (RoboFlow 100) datasets available in torchvision.
+#' This data frame contains metadata about each dataset including descriptions, sizes,
+#' available splits, and collection information.
+#'
+#' @format A data frame with datasets as rows and 16 columns:
+#' \describe{
+#'   \item{collection}{Collection name (biology, medical, infrared, damage, underwater, document)}
+#'   \item{dataset}{Dataset identifier used in collection functions}
+#'   \item{description}{Brief description of the dataset and its purpose}
+#'   \item{task}{Machine learning task type (currently all "object_detection")}
+#'   \item{num_images}{Total number of images across all splits}
+#'   \item{image_width}{Typical image width in pixels}
+#'   \item{image_height}{Typical image height in pixels}
+#'   \item{train_size_mb}{Size of training split in megabytes}
+#'   \item{test_size_mb}{Size of test split in megabytes}
+#'   \item{valid_size_mb}{Size of validation split in megabytes}
+#'   \item{total_size_mb}{Total size across all splits in megabytes}
+#'   \item{has_train}{Logical indicating if training split is available}
+#'   \item{has_test}{Logical indicating if test split is available}
+#'   \item{has_valid}{Logical indicating if validation split is available}
+#'   \item{function_name}{R function name to load this dataset's collection}
+#'   \item{roboflow_url}{URL to the collection on RoboFlow Universe}
+#' }
+#'
+#' @examples
+#' \dontrun{
+#' # View the complete catalog
+#' data(rf100_catalog)
+#' View(rf100_catalog)
+#'
+#' # See all biology datasets
+#' subset(rf100_catalog, collection == "biology")
+#'
+#' # Find large datasets (> 100 MB)
+#' subset(rf100_catalog, total_size_mb > 100)
+#' }
+#'
+#' @seealso [search_rf100()], [get_rf100_catalog()]
+"rf100_catalog"
+
+#' Search RF100 Dataset Catalog
+#'
+#' Search through all RF100 datasets by keywords in name or description,
+#' or filter by collection. This makes it easy to discover datasets relevant
+#' to your task without browsing each collection individually.
+#'
+#' @param keyword Character string to search for (case-insensitive). Searches
+#'   in both dataset names and descriptions. If NULL, returns all datasets
+#'   (optionally filtered by collection).
+#' @param collection Filter by collection name. One of: "biology", "medical",
+#'   "infrared", "damage", "underwater", "document". If NULL, searches all collections.
+#'
+#' @return A data frame with matching datasets and their metadata. Returns NULL
+#'   invisibly if no matches are found.
+#'
+#' @examples
+#' \dontrun{
+#' # Find all medical datasets
+#' search_rf100(collection = "medical")
+#'
+#' # Find datasets about cells
+#' search_rf100("cell")
+#'
+#' # Find photovoltaic/solar datasets
+#' search_rf100("solar")
+#' search_rf100("photovoltaic")
+#'
+#' # Find all biology datasets with "cell" in name/description
+#' search_rf100("cell", collection = "biology")
+#'
+#' # List all available datasets
+#' search_rf100()
+#' }
+#'
+#' @seealso [get_rf100_catalog()], [rf100_catalog]
+#' @export
+search_rf100 <- function(keyword = NULL, collection = NULL) {
+  # Load the catalog
+  utils::data("rf100_catalog", package = "torchvision", envir = environment())
+
+  result <- rf100_catalog
+
+  # Filter by collection first
+  if (!is.null(collection)) {
+    valid_collections <- c("biology", "medical", "infrared", "damage", "underwater", "document")
+    if (!collection %in% valid_collections) {
+      stop("Invalid collection. Must be one of: ", paste(valid_collections, collapse = ", "))
+    }
+    result <- result[result$collection == collection, ]
+  }
+
+  # Filter by keyword in dataset name or description
+  if (!is.null(keyword)) {
+    pattern <- tolower(keyword)
+    matches <- grepl(pattern, tolower(result$dataset)) |
+               grepl(pattern, tolower(result$description))
+    result <- result[matches, ]
+  }
+
+  if (nrow(result) == 0) {
+    message("No datasets found matching criteria")
+    return(invisible(NULL))
+  }
+
+  # Reset row names for cleaner display
+  rownames(result) <- NULL
+
+  result
+}
+
+#' Get Complete RF100 Catalog
+#'
+#' Returns the complete catalog of all RF100 datasets with their metadata.
+#' This is a convenience function that loads and returns the rf100_catalog data.
+#'
+#' @return A data frame with all RF100 datasets and their metadata.
+#'
+#' @examples
+#' \dontrun{
+#' # Get complete catalog
+#' catalog <- get_rf100_catalog()
+#'
+#' # View in RStudio
+#' View(catalog)
+#'
+#' # Summary statistics
+#' summary(catalog$total_size_mb)
+#' table(catalog$collection)
+#'
+#' # Find smallest dataset
+#' catalog[which.min(catalog$total_size_mb), ]
+#'
+#' # Find largest dataset
+#' catalog[which.max(catalog$total_size_mb), ]
+#' }
+#'
+#' @seealso [search_rf100()], [rf100_catalog]
+#' @export
+get_rf100_catalog <- function() {
+  utils::data("rf100_catalog", package = "torchvision", envir = environment())
+  rf100_catalog
+}
+
+#' List Datasets in an RF100 Collection
+#'
+#' List all available datasets within a specific RF100 collection.
+#'
+#' @param collection Collection name. One of: "biology", "medical", "infrared",
+#'   "damage", "underwater", "document".
+#'
+#' @return Character vector of dataset names in the collection.
+#'
+#' @examples
+#' \dontrun{
+#' # List all biology datasets
+#' list_rf100_datasets("biology")
+#'
+#' # List all medical datasets
+#' list_rf100_datasets("medical")
+#' }
+#'
+#' @seealso [search_rf100()], [get_rf100_catalog()]
+#' @export
+list_rf100_datasets <- function(collection) {
+  valid_collections <- c("biology", "medical", "infrared", "damage", "underwater", "document")
+  if (!collection %in% valid_collections) {
+    stop("Invalid collection. Must be one of: ", paste(valid_collections, collapse = ", "))
+  }
+
+  utils::data("rf100_catalog", package = "torchvision", envir = environment())
+  datasets <- rf100_catalog[rf100_catalog$collection == collection, "dataset"]
+  sort(datasets)
+}
+

README.md

@@ -23,3 +23,54 @@ You can install the development version from GitHub with:
 remotes::install_github("mlverse/torchvision@main")
 ```
 
+## RF100 Dataset Catalog
+
+torchvision includes 34 datasets from the RoboFlow 100 benchmark, organized into 6 collections. Use the catalog to easily discover and search for datasets:
+
+```r
+library(torchvision)
+
+# Search for datasets by keyword
+search_rf100("solar")       # Find solar/photovoltaic datasets
+search_rf100("cell")        # Find cell-related datasets
+search_rf100("medical")     # Find medical imaging datasets
+
+# Browse by collection
+search_rf100(collection = "biology")    # All biology datasets
+search_rf100(collection = "medical")    # All medical datasets
+
+# View complete catalog
+catalog <- get_rf100_catalog()
+View(catalog)
+```
+
+### Available Collections
+
+- **Biology** (9 datasets): Microscopy, cells, bacteria, parasites, plant diseases
+- **Medical** (8 datasets): X-rays, MRI, pathology, tumor detection
+- **Infrared** (4 datasets): Thermal imaging, FLIR cameras, solar panels
+- **Damage** (3 datasets): Infrastructure damage, defect detection
+- **Underwater** (4 datasets): Marine life, coral reefs, underwater objects
+- **Document** (6 datasets): OCR, document parsing, diagrams
+
+See `vignette("rf100-datasets")` for the complete catalog and detailed information.
+
+### Example Usage
+
+```r
+# Search for a dataset
+search_rf100("blood")
+
+# Load the dataset
+ds <- rf100_biology_collection(
+  dataset = "blood_cell",
+  split = "train",
+  download = TRUE
+)
+
+# Visualize a sample
+item <- ds[1]
+boxed <- draw_bounding_boxes(item)
+tensor_image_browse(boxed)
+```
+