Add image-captioning tutorial (#59)

Author: mfl28

.gitignore

@@ -33,7 +33,6 @@
 
 # PyTorch
 libtorch/
-build/
 
 # Datasets
 data/
@@ -56,6 +55,8 @@ install_manifest.txt
 compile_commands.json
 CTestTestfile.cmake
 _deps
+build/
+cmake-build-*/
 
 # Other
 .DS_Store
@@ -66,4 +67,4 @@ _deps
 
 # extern
 extern/*
-!extern/CMakeLists.txt
+!extern/CMakeLists.txt

CMakeLists.txt

@@ -57,6 +57,7 @@ add_subdirectory("tutorials/intermediate/language_model")
 add_subdirectory("tutorials/advanced/generative_adversarial_network")
 add_subdirectory("tutorials/advanced/variational_autoencoder")
 add_subdirectory("tutorials/advanced/neural_style_transfer")
+add_subdirectory("tutorials/advanced/image_captioning")
 
 if(MSVC)
     include(copy_torch_dlls)

README.md

@@ -120,7 +120,7 @@ cmake --build build
 * [Generative Adversarial Networks](tutorials/advanced/generative_adversarial_network/main.cpp)
 * [Variational Auto-Encoder](tutorials/advanced/variational_autoencoder/src/main.cpp)
 * [Neural Style Transfer](tutorials/advanced/neural_style_transfer/src/main.cpp)
-* [Image Captioning (CNN-RNN)]()
+* [Image Captioning (CNN-AttentionRNN)](tutorials/advanced/image_captioning/src/main.cpp)
 
 ### 4. Interactive Tutorials
 * [Tensor Slicing](notebooks/tensor_slicing.ipynb)

cmake/cpplint.cmake

@@ -21,8 +21,8 @@ endif()
 execute_process(COMMAND ${CPPLINT_COMMAND}
                 "--linelength=120" 
                 "--recursive" 
-                "--filter=-build/include_subdir,-build/include_what_you_use"
+                "--filter=-build/include_subdir,-build/include_what_you_use,-build/c++11,-runtime/references"
                 "main.cpp"
                 "tutorials"
                 "utils"
-                WORKING_DIRECTORY "${CMAKE_CURRENT_LIST_DIR}/..")
+                WORKING_DIRECTORY "${CMAKE_CURRENT_LIST_DIR}/..")

cmake/download_datasets.cmake

@@ -5,4 +5,5 @@ include(download_mnist)
 include(download_cifar10)
 include(download_penntreebank)
 include(download_neural_style_transfer_images)
+include(download_flickr8k)
 message(STATUS "Fetching datasets - done")

cmake/download_flickr8k.cmake

@@ -0,0 +1,39 @@
+cmake_minimum_required(VERSION 3.14 FATAL_ERROR)
+
+include(FetchContent)    
+
+set(FLICKR8K_DIR "${CMAKE_SOURCE_DIR}/data/flickr_8k")
+
+set(FLICKR8K_DATA_URL "https://github.com/jbrownlee/Datasets/releases/download/Flickr8k/Flickr8k_Dataset.zip")
+set(FLICKR8K_TEXT_URL "https://github.com/jbrownlee/Datasets/releases/download/Flickr8k/Flickr8k_text.zip")
+
+set(FLICKR8k_DATA_SOURCE_DIR "${FLICKR8K_DIR}/Flickr8k_Dataset")
+set(FLICKR8k_TEXT_SOURCE_DIR "${FLICKR8K_DIR}/Flickr8k_text")
+
+if(NOT EXISTS ${FLICKR8k_TEXT_SOURCE_DIR} OR NOT EXISTS ${FLICKR8k_DATA_SOURCE_DIR})
+    message(STATUS "Fetching Flickr8k dataset...")
+    
+    FetchContent_Declare(
+        flickr_8k_text
+        DOWNLOAD_DIR ${FLICKR8K_DIR}/download
+        SOURCE_DIR ${FLICKR8k_TEXT_SOURCE_DIR}
+        URL ${FLICKR8K_TEXT_URL}
+    )
+
+    FetchContent_MakeAvailable(flickr_8k_text)
+
+    FetchContent_Declare(
+        flickr_8k_data
+        DOWNLOAD_DIR ${FLICKR8K_DIR}/download
+        SOURCE_DIR ${FLICKR8k_DATA_SOURCE_DIR}
+        URL ${FLICKR8K_DATA_URL}
+    )
+
+    FetchContent_MakeAvailable(flickr_8k_data)
+
+    file(REMOVE_RECURSE "${FLICKR8K_DIR}/download")
+    
+    message(STATUS "Fetching Flickr8k dataset - done")
+else()
+    message(STATUS "Flickr8k dataset already present, skipping...")
+endif()

cmake/download_libtorch.cmake

@@ -53,4 +53,4 @@ FetchContent_MakeAvailable(libtorch)
 
 message(STATUS "Downloading libtorch - done")
 
-find_package(Torch REQUIRED PATHS "${CMAKE_SOURCE_DIR}/libtorch")
+find_package(Torch REQUIRED PATHS "${CMAKE_SOURCE_DIR}/libtorch")