Add PR suggestions

examples/fit-model-ceres.cpp

@@ -303,10 +303,8 @@ int main(int argc, char* argv[])
     // These will be the 2D image points and their corresponding 3D vertex id's used for the fitting
     auto& landmarks = fitting_data.landmarks;
 
-    auto landmarks_definitions = morphable_model.get_landmark_definitions();
-    auto* landmarks_definitions_ptr =
-        landmarks_definitions.has_value() ? &landmarks_definitions.value() : nullptr;
-    auto indexed_landmarks = landmark_mapper.get_indexed_landmarks(landmarks, landmarks_definitions_ptr);
+    auto indexed_landmarks =
+        landmark_mapper.get_indexed_landmarks(landmarks, morphable_model.get_landmark_definitions());
 
     google::InitGoogleLogging(argv[0]); // Ceres logging initialisation
     std::stringstream fitting_log;
@@ -319,9 +317,9 @@ int main(int argc, char* argv[])
                              EOS_CERES_EXAMPLE_COLOR_COEFFS_NUM>(&morphable_model, &blendshapes);
     auto camera = fitting::PerspectiveCameraParameters(image.width(), image.height());
     model_fitter.add_camera_cost_function(camera, indexed_landmarks);
-    model_fitter.block_shapes_fitting();
-    model_fitter.block_blendshapes_fitting();
-    model_fitter.block_fov_fitting(camera, 60.0);
+    model_fitter.set_shape_coefficients_constant();
+    model_fitter.set_blendshape_coefficients_constant();
+    model_fitter.set_fov_constant(camera, 60.0);
 
     auto solver_summary = model_fitter.solve();
 
@@ -339,7 +337,7 @@ int main(int argc, char* argv[])
                                       camera.calculate_projection_matrix(), camera.get_viewport());
     ceres_example::draw_landmarks(outimg, indexed_landmarks);
 
-    auto camera_euler_rotation = camera.get_euler_rotation();
+    const auto camera_euler_rotation = camera.get_euler_rotation();
     fitting_log << "Pose fit with mean shape:\tYaw " << glm::degrees(camera_euler_rotation[1]) << ", Pitch "
                 << glm::degrees(camera_euler_rotation[0]) << ", Roll "
                 << glm::degrees(camera_euler_rotation[2]) << "; t & f: " << camera.translation_and_intrinsics
@@ -368,7 +366,7 @@ int main(int argc, char* argv[])
 
     model_fitter.reset_problem();
     model_fitter.add_camera_cost_function(camera, indexed_landmarks);
-    model_fitter.block_fov_fitting(camera, 60.0);
+    model_fitter.set_fov_constant(camera, 60.0);
 
     model_fitter.add_shape_prior_cost_function();
     model_fitter.add_blendshape_prior_cost_function();
@@ -398,7 +396,7 @@ int main(int argc, char* argv[])
     // Draw the landmarks projected using all estimated parameters:
     // Construct the rotation & translation (model-view) matrices, projection matrix, and viewport:
 
-    auto points = model_fitter.calculate_estimated_points_positions();
+    const auto points = model_fitter.calculate_estimated_points_positions();
 
     core::Mesh mesh = morphablemodel::sample_to_mesh(
         points, color_instance, morphable_model.get_shape_model().get_triangle_list(),

include/eos/core/LandmarkMapper.hpp

@@ -103,8 +103,7 @@ class LandmarkMapper
      *
      * @param[in] mappings A set of landmark mappings.
      */
-    LandmarkMapper(std::unordered_map<std::string, std::string> mappings)
-        : landmark_mappings(std::move(mappings)){};
+    LandmarkMapper(std::unordered_map<std::string, std::string>& mappings) : landmark_mappings(mappings){};
 
     /**
      * @brief Converts the given landmark name to the mapped name.
@@ -141,7 +140,7 @@ class LandmarkMapper
      */
     template <typename LandmarkType>
     auto get_indexed_landmarks(const LandmarkCollection<LandmarkType>& landmarks,
-                               const LandmarkDefinitions* const landmark_definitions = nullptr) const
+                               const cpp17::optional<LandmarkDefinitions> landmark_definitions = cpp17::nullopt) const
     {
         IndexedLandmarkCollection<LandmarkType> indexed_landmarks;
         for (auto& landmark : landmarks)
@@ -153,7 +152,7 @@ class LandmarkMapper
             }
 
             int vertex_idx;
-            if (landmark_definitions != nullptr)
+            if (landmark_definitions)
             {
                 const auto found_vertex_idx = landmark_definitions->find(converted_name.value());
 

include/eos/fitting/ceres_nonlinear.hpp

@@ -161,19 +161,18 @@ struct LandmarkCost
                                                   blendshape_coeffs, get_num_cam_params(use_perspective));
 
         // Project the point to 2D:
-        const auto point_3d = tvec3<T>(point_arr[0], point_arr[1], point_arr[2]);
+        const tvec3<T> point_3d(point_arr[0], point_arr[1], point_arr[2]);
         // I think the quaternion is always normalised because we run Ceres with QuaternionParameterization
-        const auto rot_quat =
-            tquat<T>(camera_rotation[0], camera_rotation[1], camera_rotation[2], camera_rotation[3]);
+        const tquat<T> rot_quat(camera_rotation[0], camera_rotation[1], camera_rotation[2], camera_rotation[3]);
         // We rotate ZXY*p, which is RPY*p. I'm not sure this matrix still corresponds to RPY - probably if we
         // use glm::eulerAngles(), these are not RPY anymore and we'd have to adjust if we were to use
         // rotation matrices.
         const auto rot_mtx = glm::mat4_cast(rot_quat);
 
         // Todo: use get_opencv_viewport() from nonlin_cam_esti.hpp.
-        const auto viewport = tvec4<T>(0, image_height, image_width, -image_height); // OpenCV convention
+        const tvec4<T> viewport(0, image_height, image_width, -image_height); // OpenCV convention
 
-        auto projected_point = tvec3<T>(); // Note: could avoid default construction by using a lambda and
+        tvec3<T> projected_point(); // Note: could avoid default construction by using a lambda and
                                            // immediate invocation
         if (use_perspective)
         {
@@ -464,7 +463,6 @@ ceres::Solver::Options get_default_solver_options()
     ceres::Solver::Options solver_options;
     solver_options.linear_solver_type = ceres::ITERATIVE_SCHUR;
     solver_options.num_threads = 8;
-    solver_options.num_linear_solver_threads = 8; // only SPARSE_SCHUR can use this
     solver_options.minimizer_progress_to_stdout = true;
     solver_options.max_num_iterations = 50;
 
@@ -612,9 +610,24 @@ class ModelFitter
      */
     explicit ModelFitter(const morphablemodel::MorphableModel* const morphable_model,
                          const morphablemodel::Blendshapes* const blendshapes = nullptr)
-        : problem(std::make_unique<ceres::Problem>()), morphable_model(morphable_model),
-          blendshapes(find_blendshapes(blendshapes))
+        : problem(std::make_unique<ceres::Problem>()), morphable_model(morphable_model)
     {
+        if (blendshapes == nullptr)
+        {
+            if (!morphable_model->has_separate_expression_model())
+            {
+                throw std::runtime_error(
+                    "Blendshapes was not passed and morphable model does not contain them too."
+                    "Try to pass blendshapes explicitly.");
+            } else
+            {
+                this->blendshapes =
+                    &cpp17::get<morphablemodel::Blendshapes>(morphable_model->get_expression_model().value());
+            }
+        } else
+        {
+            this->blendshapes = blendshapes;
+        }
     }
 
     /*
@@ -820,15 +833,15 @@ class ModelFitter
     /**
      * Block shapes fitting
      */
-    void block_shapes_fitting()
+    void set_shape_coefficients_constant()
     {
         problem->SetParameterBlockConstant(&shape_coefficients[0]);
     }
 
     /**
      * Block blendshapes fitting
      */
-    void block_blendshapes_fitting()
+    void set_blendshape_coefficients_constant()
     {
         problem->SetParameterBlockConstant(&blendshape_coefficients[0]);
     }
@@ -841,7 +854,7 @@ class ModelFitter
      * @param[in,out] camera perspective camera to fit
      * @param[in] fov_in_grad field of view to fix
      */
-    void block_fov_fitting(PerspectiveCameraParameters& camera, double fov_in_grad)
+    void set_fov_constant(PerspectiveCameraParameters &camera, double fov_in_grad)
     {
         auto& camera_translation_and_intrinsics = camera.translation_and_intrinsics;
         camera_translation_and_intrinsics[3] = glm::radians(fov_in_grad);
@@ -859,32 +872,6 @@ class ModelFitter
     std::unique_ptr<ceres::Problem> problem;
 
 private:
-    /*
-     * Check if blendshapes is not nullptr or try to find them in morphable_model.
-     *
-     * * @return std::array with 3 or 4 (for perspective projection) parameters.
-     */
-    auto find_blendshapes(const morphablemodel::Blendshapes* const blendshapes = nullptr) const
-    {
-        if (blendshapes == nullptr)
-        {
-            const auto& blendshapes_optional = morphable_model->get_expression_model();
-
-            if (blendshapes_optional.has_value())
-            {
-                return &cpp17::get<morphablemodel::Blendshapes>(*blendshapes_optional);
-            } else
-            {
-                throw std::runtime_error(
-                    "Blendshapes was not passed and morphable model does not contain them too."
-                    "Try to pass blendshapes explicitly.");
-            }
-        } else
-        {
-            return blendshapes;
-        }
-    }
-
     const morphablemodel::MorphableModel* const morphable_model;
     const morphablemodel::Blendshapes* const blendshapes;
 };

include/eos/fitting/contour_correspondence.hpp

@@ -78,7 +78,7 @@ struct ModelContour
 
     // We store r/l separately because we currently only fit to the contour facing the camera.
     // Also if we were to fit to the whole contour: Be careful not to just fit to the closest. The
-    // "invisible" ones behind might be closer on an e.g 90° angle. Store CNT for left/right side separately?
+    // "invisible" ones behind might be closer on an e.g 90° angle. Store CNT for left/right side separately?
 
     /**
      * Helper method to load a ModelContour from
@@ -265,7 +265,7 @@ get_contour_correspondences(const core::LandmarkCollection<Eigen::Vector2f>& lan
  * have different size. 
  * Correspondence can be established using get_nearest_contour_correspondences().
  *
- * If the yaw angle is between +-7.5°, both contours will be selected.
+ * If the yaw angle is between +-7.5°, both contours will be selected.
  *
  * @param[in] yaw_angle Yaw angle in degrees.
  * @param[in] contour_landmarks 2D image contour ids of left or right side (for example for ibug landmarks).
@@ -298,7 +298,7 @@ select_contour(float yaw_angle, const ContourLandmarks& contour_landmarks, const
                                             begin(contour_landmarks.left_contour),
                                             end(contour_landmarks.left_contour));
     }
-    // Note there's an overlap between the angles - if a subject is between +- 7.5°, both contours get added.
+    // Note there's an overlap between the angles - if a subject is between +- 7.5°, both contours get added.
     return std::make_pair(contour_landmark_identifiers, model_contour_indices);
 };