cli: add model search path

CMakeLists.txt

@@ -1,15 +1,25 @@
 cmake_minimum_required(VERSION 3.28)
 
-project(vision.cpp VERSION 0.3.0 LANGUAGES CXX)
+project(vision.cpp VERSION 0.3.1 LANGUAGES CXX)
 
 option(BUILD_SHARED_LIBS "Build shared libraries instead of static libraries" ON)
 option(VISP_VULKAN "Enable Vulkan support" OFF)
 option(VISP_DEV "Enable development mode" OFF)
 option(VISP_CI "Enable for continuous integration environment" OFF)
 option(VISP_TESTS "Build tests" ${PROJECT_IS_TOP_LEVEL})
+option(VISP_INSTALL_MODELS "Download and install default models" OFF)
 option(VISP_FMT_LIB "Use external fmt library instead of standard C++ <format>" OFF)
+option(VISP_STATIC_GGML "Force static build of ggml" OFF)
 option(VISP_ASAN "Enable AddressSanitizer" OFF)
 
+include(GNUInstallDirs)
+set(VISP_INCLUDE_INSTALL_DIR ${CMAKE_INSTALL_INCLUDEDIR} CACHE PATH "Location of header files")
+set(VISP_LIB_INSTALL_DIR ${CMAKE_INSTALL_LIBDIR} CACHE PATH "Location of library files")
+set(VISP_BIN_INSTALL_DIR ${CMAKE_INSTALL_BINDIR} CACHE PATH "Location of binary files")
+if(VISP_INSTALL_MODELS)
+  set(VISP_MODEL_INSTALL_DIR "${CMAKE_INSTALL_DATAROOTDIR}/visioncpp" CACHE PATH "Directory to install default models to")
+endif()
+
 if(PROJECT_IS_TOP_LEVEL)
   set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
   set(CMAKE_ARCHIVE_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/lib)
@@ -94,6 +104,12 @@ if(VISP_FMT_LIB)
   list(APPEND VISP_DEFINITIONS VISP_FMT_LIB)
 endif()
 
+set(BUILD_SHARED_LIBS_SAVED ${BUILD_SHARED_LIBS})
+if(VISP_STATIC_GGML)
+  set(BUILD_SHARED_LIBS OFF)
+  set(CMAKE_POSITION_INDEPENDENT_CODE ON)
+  set(GGML_STATIC ON)
+endif()
 set(GGML_VULKAN ${VISP_VULKAN})
 set(GGML_LLAMAFILE ON)
 if(VISP_CI)
@@ -110,6 +126,7 @@ if(VISP_CI)
   endif()
 endif()
 add_subdirectory(depend/llama/ggml)
+set(BUILD_SHARED_LIBS ${BUILD_SHARED_LIBS_SAVED})
 
 # Vision.cpp libraries and executables
 
@@ -121,15 +138,13 @@ add_subdirectory(src/cli)
 if(VISP_TESTS)
   enable_testing()
   add_subdirectory(tests)
+endif()
+if(VISP_TESTS OR VISP_INSTALL_MODELS)
   add_subdirectory(models)
 endif()
 
 # Installation
 
-set(VISP_INCLUDE_INSTALL_DIR ${CMAKE_INSTALL_INCLUDEDIR} CACHE PATH "Location of header files")
-set(VISP_LIB_INSTALL_DIR ${CMAKE_INSTALL_LIBDIR} CACHE PATH "Location of library files")
-set(VISP_BIN_INSTALL_DIR ${CMAKE_INSTALL_BINDIR} CACHE PATH "Location of binary files")
-
 install(TARGETS visioncpp
   RUNTIME DESTINATION ${VISP_BIN_INSTALL_DIR}
   LIBRARY DESTINATION ${VISP_LIB_INSTALL_DIR}
@@ -138,6 +153,9 @@ install(DIRECTORY include/ DESTINATION ${VISP_INCLUDE_INSTALL_DIR})
 if(PROJECT_IS_TOP_LEVEL)
   install(FILES README.md LICENSE DESTINATION .)
 endif()
+if(VISP_INSTALL_MODELS)
+  install(FILES ${VISP_DEFAULT_MODELS} DESTINATION ${VISP_MODEL_INSTALL_DIR})
+endif()
 
 install(TARGETS vision-cli RUNTIME DESTINATION ${VISP_BIN_INSTALL_DIR})
 

models/CMakeLists.txt

@@ -1,37 +1,43 @@
-# Download models used in tests (happens only if VISP_TESTS is enabled)
+# Download models used in tests or for installation
 
-message(STATUS "Checking for models/MobileSAM-F16.gguf")
-file(DOWNLOAD
+function(download_model url filename hash)
+  message(STATUS "Checking for ${filename}")
+  file(DOWNLOAD
+    ${url}
+    ${CMAKE_CURRENT_LIST_DIR}/${filename}
+    EXPECTED_HASH ${hash}
+    SHOW_PROGRESS
+  )
+  set(DEFAULT_MODELS "${DEFAULT_MODELS};${CMAKE_CURRENT_LIST_DIR}/${filename}" PARENT_SCOPE)
+endfunction()
+
+
+download_model(
   "https://huggingface.co/Acly/MobileSAM-GGUF/resolve/main/MobileSAM-F16.gguf"
-  ${CMAKE_CURRENT_LIST_DIR}/MobileSAM-F16.gguf
-  EXPECTED_HASH "SHA256=b546366475e3ad744bb2eaf7634df88e9aaf25f6622797d2de300f5a530831f7"
-  SHOW_PROGRESS
+  "MobileSAM-F16.gguf"
+  "SHA256=b546366475e3ad744bb2eaf7634df88e9aaf25f6622797d2de300f5a530831f7"
 )
-message(STATUS "Checking for models/BiRefNet-lite-F16.gguf")
-file(DOWNLOAD
+download_model(
   "https://huggingface.co/Acly/BiRefNet-GGUF/resolve/main/BiRefNet-lite-F16.gguf"
-  ${CMAKE_CURRENT_LIST_DIR}/BiRefNet-lite-F16.gguf
-  EXPECTED_HASH "SHA256=7b5397a2c98d66677f8f74317774bbeac49dbb321b8a3dc744af913db71d4fa5"
-  SHOW_PROGRESS
+  "BiRefNet-lite-F16.gguf"
+  "SHA256=7b5397a2c98d66677f8f74317774bbeac49dbb321b8a3dc744af913db71d4fa5"
 )
-message(STATUS "Checking for models/Depth-Anything-V2-Small-F16.gguf")
-file(DOWNLOAD
+download_model(
   "https://huggingface.co/Acly/Depth-Anything-V2-GGUF/resolve/main/Depth-Anything-V2-Small-F16.gguf"
-  ${CMAKE_CURRENT_LIST_DIR}/Depth-Anything-V2-Small-F16.gguf
-  EXPECTED_HASH "SHA256=0f83332d6a8b4375cd7fdcc168f3e3636f474f8e84b0959e903f513aace782f5"
-  SHOW_PROGRESS
+  "Depth-Anything-V2-Small-F16.gguf"
+  "SHA256=0f83332d6a8b4375cd7fdcc168f3e3636f474f8e84b0959e903f513aace782f5"
 )
-message(STATUS "Checking for models/MIGAN-512-places2-F16.gguf")
-file(DOWNLOAD
+download_model(
   "https://huggingface.co/Acly/MIGAN-GGUF/resolve/main/MIGAN-512-places2-F16.gguf"
-  ${CMAKE_CURRENT_LIST_DIR}/MIGAN-512-places2-F16.gguf
-  EXPECTED_HASH "SHA256=3e47592bf716d0dc306f8dc02d4476cfcdaf2c055fa3c3c8e0ced4db775eb64b"
-  SHOW_PROGRESS
+  "MIGAN-512-places2-F16.gguf"
+  "SHA256=3e47592bf716d0dc306f8dc02d4476cfcdaf2c055fa3c3c8e0ced4db775eb64b"
 )
-message(STATUS "Checking for models/RealESRGAN-x4plus_anime-6B-F16.gguf")
-file(DOWNLOAD
-  "https://huggingface.co/Acly/Real-ESRGAN-GGUF/resolve/main/RealESRGAN-x4plus_anime-6B-F16.gguf"
-  ${CMAKE_CURRENT_LIST_DIR}/RealESRGAN-x4plus_anime-6B-F16.gguf
-  EXPECTED_HASH "SHA256=730469c5a2269cdef96d0d58aacf87bcf25d7a0d92256685808e6cdce0675c09"
-  SHOW_PROGRESS
-)
+if(VISP_TESTS)
+  download_model(
+    "https://huggingface.co/Acly/Real-ESRGAN-GGUF/resolve/main/RealESRGAN-x4plus_anime-6B-F16.gguf"
+    "RealESRGAN-x4plus_anime-6B-F16.gguf"
+    "SHA256=730469c5a2269cdef96d0d58aacf87bcf25d7a0d92256685808e6cdce0675c09"
+  )
+endif()
+
+set(VISP_DEFAULT_MODELS "${DEFAULT_MODELS}" PARENT_SCOPE)

src/cli/CMakeLists.txt

@@ -4,4 +4,13 @@ target_include_directories(vision-cli PRIVATE ..)
 target_compile_definitions(vision-cli PRIVATE ${VISP_ASSERT} ${VISP_DEFINITIONS})
 target_compile_options(vision-cli PRIVATE ${VISP_WARNINGS} ${VISP_COMP_OPTIONS})
 target_link_options(vision-cli PRIVATE ${VISP_LINK_OPTIONS})
-target_link_libraries(vision-cli PRIVATE ggml visioncpp ${VISP_FMT_LINK})
+target_link_libraries(vision-cli PRIVATE ggml visioncpp ${VISP_FMT_LINK})
+
+if(VISP_INSTALL_MODELS)
+  if(IS_ABSOLUTE "${VISP_MODEL_INSTALL_DIR}")
+    set(MODEL_DIR "${VISP_MODEL_INSTALL_DIR}")
+  else()
+    set(MODEL_DIR "${CMAKE_INSTALL_PREFIX}/${VISP_MODEL_INSTALL_DIR}")
+  endif()
+endif()
+target_compile_definitions(vision-cli PRIVATE VISP_MODEL_INSTALL_DIR="${MODEL_DIR}")

src/cli/cli.cpp

@@ -140,7 +140,7 @@ cli_args cli_parse(int argc, char** argv) {
         } else if (arg == "-o" || arg == "--output") {
             r.output = next_arg(argc, argv, i);
         } else if (arg == "-m" || arg == "--model") {
-            r.model = validate_path(next_arg(argc, argv, i));
+            r.model = next_arg(argc, argv, i);
         } else if (arg == "-p" || arg == "--prompt") {
             r.prompt = collect_args(argc, argv, i, '-');
         } else if (arg == "-b" || arg == "--backend") {
@@ -245,6 +245,42 @@ char const* to_string(tensor_data_layout l) {
     }
 }
 
+path find_model(char const* model_name_or_path) {
+    path p = path(model_name_or_path);
+    if (exists(p) || p.is_absolute()) {
+        return p;
+    }
+    path search_paths[5];
+    search_paths[0] = path("models");
+    if (char const* vision_model_dir = getenv("VISION_MODEL_DIR")) {
+        search_paths[1] = path(vision_model_dir);
+    }
+    if (char const* xdg_data_home = getenv("XDG_DATA_HOME")) {
+        search_paths[2] = path(xdg_data_home) / "visioncpp";
+    }
+    if (char const* home = getenv("HOME")) {
+        search_paths[3] = path(home) / ".local/share/visioncpp";
+    }
+    if constexpr (VISP_MODEL_INSTALL_DIR[0] != '\0') {
+        search_paths[4] = path(VISP_MODEL_INSTALL_DIR);
+    }
+    for (auto& sp : search_paths) {
+        if (!sp.empty()) {
+            path candidate = sp / p;
+            if (exists(candidate)) {
+                return candidate;
+            }
+        }
+    }
+    printf("Looking for %s\n", p.generic_string().c_str());
+    for (auto& sp : search_paths) {
+        if (!sp.empty()) {
+            printf("Looking for %s\n", (sp / p).generic_string().c_str());
+        }
+    }
+    throw except("Model file not found: {}", model_name_or_path);
+}
+
 std::tuple<model_file, model_weights> load_model_weights(
     cli_args const& args,
     backend_device const& dev,
@@ -253,10 +289,11 @@ std::tuple<model_file, model_weights> load_model_weights(
     tensor_data_layout preferred_layout = tensor_data_layout::unknown) {
 
     timer t;
-    char const* model_path = args.model ? args.model : default_model;
-    printf("Loading model weights from '%s'... ", model_path);
+    path model_path = find_model(args.model ? args.model : default_model);
+    auto model_path_str = model_path.generic_string();
+    printf("Loading model weights from '%s'... ", model_path_str.c_str());
 
-    model_file file = model_load(model_path);
+    model_file file = model_load(model_path_str.c_str());
     model_weights weights = model_init(file.n_tensors() + n_tensors);
     if (preferred_layout == tensor_data_layout::unknown) {
         preferred_layout = file.tensor_layout();
@@ -355,7 +392,7 @@ sam_prompt sam_parse_prompt(std::span<char const* const> args, i32x2 extent) {
 void run_sam(cli_args const& args) {
     backend_device backend = backend_init(args);
     auto [file, weights] = load_model_weights(
-        args, backend, "models/MobileSAM-F16.gguf", 0, backend.preferred_layout());
+        args, backend, "MobileSAM-F16.gguf", 0, backend.preferred_layout());
     sam_params params{};
 
     require_inputs(args.inputs, 1, "<image>");
@@ -409,7 +446,7 @@ void run_sam(cli_args const& args) {
 void run_birefnet(cli_args const& args) {
     backend_device backend = backend_init(args);
     auto [file, weights] = load_model_weights(
-        args, backend, "models/BiRefNet-F16.gguf", 0, backend.preferred_layout());
+        args, backend, "BiRefNet-lite-F16.gguf", 0, backend.preferred_layout());
 
     require_inputs(args.inputs, 1, "<image>");
     image_data image = image_load(args.inputs[0]);
@@ -453,7 +490,7 @@ void run_birefnet(cli_args const& args) {
 void run_depth_anything(cli_args const& args) {
     backend_device backend = backend_init(args);
     auto [file, weights] = load_model_weights(
-        args, backend, "models/DepthAnythingV2-Small-F32.gguf", 0, backend.preferred_layout());
+        args, backend, "DepthAnythingV2-Small-F32.gguf", 0, backend.preferred_layout());
 
     require_inputs(args.inputs, 1, "<image>");
     image_data image = image_load(args.inputs[0]);
@@ -489,7 +526,7 @@ void run_depth_anything(cli_args const& args) {
 void run_migan(cli_args const& args) {
     backend_device backend = backend_init(args);
     auto [file, weights] = load_model_weights(
-        args, backend, "models/MIGAN-512-places2-F16.gguf", backend.preferred_layout());
+        args, backend, "MIGAN-512-places2-F16.gguf", backend.preferred_layout());
     migan_params params = migan_detect_params(file);
     params.invert_mask = true; // -> inpaint opaque areas
 
@@ -527,7 +564,7 @@ void run_migan(cli_args const& args) {
 void run_esrgan(cli_args const& args) {
     backend_device backend = backend_init(args);
     auto [file, weights] = load_model_weights(
-        args, backend, "models/RealESRGAN-x4.gguf", 0, backend.preferred_layout());
+        args, backend, "RealESRGAN-x4.gguf", 0, backend.preferred_layout());
     esrgan_params params = esrgan_detect_params(file);
     printf("- scale: %dx\n", params.scale);
     printf("- block count: %d\n", params.n_blocks);