| 165 | // ---------------------------------------------------------------------------------------- |
| 166 | |
| 167 | void bind_shape_predictors(py::module &m) |
| 168 | { |
| 169 | { |
| 170 | typedef full_object_detection type; |
| 171 | py::class_<type, std::shared_ptr<type>>(m, "full_object_detection", |
| 172 | "This object represents the location of an object in an image along with the \ |
| 173 | positions of each of its constituent parts.") |
| 174 | .def(py::init(&full_obj_det_init), py::arg("rect"), py::arg("parts"), |
| 175 | "requires \n\ |
| 176 | - rect: dlib rectangle \n\ |
| 177 | - parts: list of dlib.point, or a dlib.points object.") |
| 178 | .def_property_readonly("rect", &full_obj_det_get_rect, "Bounding box from the underlying detector. Parts can be outside box if appropriate.") |
| 179 | .def_property_readonly("num_parts", &full_obj_det_num_parts, "The number of parts of the object.") |
| 180 | .def("part", &full_obj_det_part, py::arg("idx"), "A single part of the object as a dlib point.") |
| 181 | .def("parts", &full_obj_det_parts, "A vector of dlib points representing all of the parts.") |
| 182 | .def(py::pickle(&getstate<type>, &setstate<type>)); |
| 183 | } |
| 184 | { |
| 185 | typedef shape_predictor_training_options type; |
| 186 | py::class_<type>(m, "shape_predictor_training_options", |
| 187 | "This object is a container for the options to the train_shape_predictor() routine.") |
| 188 | .def(py::init()) |
| 189 | .def_readwrite("be_verbose", &type::be_verbose, |
| 190 | "If true, train_shape_predictor() will print out a lot of information to stdout while training.") |
| 191 | .def_readwrite("cascade_depth", &type::cascade_depth, |
| 192 | "The number of cascades created to train the model with.") |
| 193 | .def_readwrite("tree_depth", &type::tree_depth, |
| 194 | "The depth of the trees used in each cascade. There are pow(2, get_tree_depth()) leaves in each tree") |
| 195 | .def_readwrite("num_trees_per_cascade_level", &type::num_trees_per_cascade_level, |
| 196 | "The number of trees created for each cascade.") |
| 197 | .def_readwrite("nu", &type::nu, |
| 198 | "The regularization parameter. Larger values of this parameter \ |
| 199 | will cause the algorithm to fit the training data better but may also \ |
| 200 | cause overfitting. The value must be in the range (0, 1].") |
| 201 | .def_readwrite("oversampling_amount", &type::oversampling_amount, |
| 202 | "The number of randomly selected initial starting points sampled for each training example") |
| 203 | .def_readwrite("oversampling_translation_jitter", &type::oversampling_translation_jitter, |
| 204 | "The amount of translation jittering to apply to bounding boxes, a good value is in in the range [0 0.5].") |
| 205 | .def_readwrite("feature_pool_size", &type::feature_pool_size, |
| 206 | "Number of pixels used to generate features for the random trees.") |
| 207 | .def_readwrite("lambda_param", &type::lambda_param, |
| 208 | "Controls how tight the feature sampling should be. Lower values enforce closer features.") |
| 209 | .def_readwrite("num_test_splits", &type::num_test_splits, |
| 210 | "Number of split features at each node to sample. The one that gives the best split is chosen.") |
| 211 | .def_readwrite("landmark_relative_padding_mode", &type::landmark_relative_padding_mode, |
| 212 | "If True then features are drawn only from the box around the landmarks, otherwise they come from the bounding box and landmarks together. See feature_pool_region_padding doc for more details.") |
| 213 | .def_readwrite("feature_pool_region_padding", &type::feature_pool_region_padding, |
| 214 | /*! |
| 215 | This algorithm works by comparing the relative intensity of pairs of |
| 216 | pixels in the input image. To decide which pixels to look at, the |
| 217 | training algorithm randomly selects pixels from a box roughly centered |
| 218 | around the object of interest. We call this box the feature pool region |
| 219 | box. |
| 220 | |
| 221 | Each object of interest is defined by a full_object_detection, which |
| 222 | contains a bounding box and a list of landmarks. If |
| 223 | landmark_relative_padding_mode==True then the feature pool region box is |
| 224 | the tightest box that contains the landmarks inside the |
no test coverage detected