Adding an article for Faster R-CNN

NEWS.md

@@ -12,6 +12,7 @@
 * Added article showcasing `model_fcn_resnet50()` with visualization utilities `draw_segmentation_masks()` and `vision_make_grid()` (@DerrickUnleashed, #281).
 * Added collection dataset catalog with `search_collection()`, `get_collection_catalog()`, and `list_collection_datasets()` functions for discovering and exploring collections (#271, @ANAMASGARD).
 * Added `target_transform_coco_masks()` and `target_transform_trimap_masks()` transformation functions for explicit segmentation mask generation (@ANAMASGARD).
+* Added article showcasing `model_fasterrcnn_resnet50_fpn()` with visualization utilities `draw_bounding_boxes()` and `vision_make_grid()` (@DerrickUnleashed, #301).
 
 ## New models
 

_pkgdown.yml

@@ -29,6 +29,8 @@ navbar:
         href: articles/examples/texture-nca.html
       - text: fcnresnet
         href: articles/examples/fcnresnet.html
+      - text: fasterrcnn
+        href: articles/examples/fasterrcnn.html
 
 reference:
 - title: Transforms

vignettes/examples/faster-rcnn.R

@@ -0,0 +1,80 @@
+# Loading Images ---------------------------------------------------
+library(torchvision)
+library(torch)
+
+url1 <- "https://raw.githubusercontent.com/pytorch/vision/main/gallery/assets/dog1.jpg"
+url2 <- "https://raw.githubusercontent.com/pytorch/vision/main/gallery/assets/dog2.jpg"
+
+dog1 <- base_loader(url1) %>% transform_to_tensor()
+dog2 <- base_loader(url2) %>% transform_to_tensor()
+
+
+# Visualizing a grid of images -------------------------------------
+
+
+dogs <- torch_stack(list(dog1, dog2))
+grid <- vision_make_grid(dogs, scale = TRUE, num_rows = 2)
+tensor_image_browse(grid)
+
+
+# Preprocessing the data -------------------------------------------
+
+
+norm_mean <- c(0.485, 0.456, 0.406)
+norm_std  <- c(0.229, 0.224, 0.225)
+
+dog1_prep <- dog1 %>%
+  transform_resize(c(800, 800)) %>%
+  transform_normalize(mean = norm_mean, std = norm_std) %>%
+  torch_tensor(dtype = torch_float32())
+dog2_prep <- dog2 %>%
+  transform_resize(c(800, 800)) %>%
+  transform_normalize(mean = norm_mean, std = norm_std) %>%
+  torch_tensor(dtype = torch_float32())
+
+# make batch (2,3,800,800)
+dog_batch <- torch_stack(list(dog1_prep, dog2_prep))
+
+
+# Loading Model ----------------------------------------------------
+
+
+model <- model_fasterrcnn_resnet50_fpn(
+  pretrained = TRUE,
+  score_thresh = 0.5,
+  nms_thresh = 0.8,
+  detections_per_img = 2
+)
+model$eval()
+
+# run model
+output <- model(dog_batch)
+
+
+# Processing the Output --------------------------------------------
+
+pred1 <- output$detections[[1]]
+pred2 <- output$detections[[2]]
+
+pred1$boxes
+pred1$labels
+pred1$scores
+
+
+# Visualizing the Output -------------------------------------------
+
+
+boxed1 <- draw_bounding_boxes(
+  dog1 %>% transform_resize(c(800, 800)),
+  boxes = pred1$boxes,
+  labels = coco_label(as.integer(pred1$labels))
+)
+
+boxed2 <- draw_bounding_boxes(
+  dog2 %>% transform_resize(c(800, 800)),
+  boxes = pred2$boxes,
+  labels = coco_label(as.integer(pred2$labels))
+)
+
+tensor_image_browse(boxed1)
+tensor_image_browse(boxed2)

vignettes/examples/faster-rcnn.Rmd

@@ -0,0 +1,9 @@
+---
+title: "faster_rcnn"
+type: docs
+---
+
+```{r, echo = FALSE}
+knitr::opts_chunk$set(eval = FALSE)
+knitr::spin_child(paste0(rmarkdown::metadata$title, ".R"))
+```