:param input_arr: :return:

:param json_file_path: :param src_dir: origin clip file path :param ori_dst_dir: :param binary_dst_dir: :param instance_dst_dir:

:param src_dir: :return:

andom color augmentation :param gt_image: :param gt_binary_image: :param gt_instance_image: :return:

Random crop image batch data for training :param gt_image: :param gt_binary_image: :param gt_instance_image: :param cropped_size:

Random horizon flip image batch data for training :param gt_image: :param gt_binary_image: :param gt_instance_image: :return:

:param image_path: :param weights_path: :param with_lane_fit: :return:

:return:

:return:

:param other: :return:

* Default constructor not supplied here */

* Remove default construction funciton */

:param args: :param kwargs: :return:

:param args: :param kwargs: :return:

:param args: :param kwargs: :return:

:param args: :param kwargs: :return:

:param args: :param kwargs: :return:

init class :param args: :param kwargs:

:param phase:

:param phase:

:param phase:

:param phase:

init lanenet backend :param phase: train or test

lane feat object :param feat: lane embeddng feats [feature_1, feature_2, ...] :param coord: lane coordinates [x, y] :

:param ipm_remap_file_path: ipm generate file path

initialize lanenet multi gpu trainner

initialize lanenet trainner

:param flags:

:return:

:param key: :param value: :return:

:return:

:param arg_value: :return:

:param gt_image: :param gt_binary_image: :param gt_instance_image: :return:

:param gt_image: :param gt_binary_image: :param gt_instance_image: :return:

* Calculate the mean feature vector among a vector of DBSCAMSample samples * @param input_samples * @return */

calculate fn figure :param input_tensor: :param label_tensor: :return:

calculate fp figure :param input_tensor: :param label_tensor: :return:

calculate accuracy acc = correct_nums / ground_truth_nums :param input_tensor: binary segmentation logits :param label_tensor: binary seg

* * @param input_samples * @param mean_feature_vec * @return */

:return:

:return:

* * @param embedding_samples * @param cluster_ret */

compute lanenet loss :param binary_seg_logits: :param binary_label: :param instance_seg_logits: :param instan

:return:

Parses an image and label from the given `serialized_example` :param serialized_example: :return:

* Detect lanes on image using lanenet model * @param input_image * @param binary_seg_result * @param pix_embedding_result */

:param input_tensor: :param k_size: :param out_dims: :param rate: :param padding: :param w_init:

:return: discriminative loss and its three components

:return:

Fully-Connected layer, takes a N>1D tensor and returns a 2D tensor. It is an equivalent of `tf.layers.dense` except for naming conven

* Gather pixel embedding features via binary segmentation result * @param binary_mask * @param pixel_embedding * @param coords * @param embedding_

Make an image summary for 4d tensor image with index idx :param img:

:param log_file_name_prefix: log文件名前缀 :return:

:param name: :param inputdata: :param data_format: :return:

:param binary_seg_logits: :param instance_seg_logits: :param name: :param reuse: :return:

:param name: :param inputdata: :param epsilon: :param data_format: :param use_affine: :return:

:return:

:return:

* Return if model is successfully initialized * @return */

:param inputdata: :param name: :param group_size: :param esp: :return:

:param name: :param inputdata: :param epsilon: epsilon to avoid divide-by-zero. :param use_bias: whether to use the e

:param inputdata: :param alpha: :param name: :return:

trim from start (in place)

trim from start (copying)

Normalize the image data by substracting the imagenet mean value :param gt_image: :param gt_binary_image: :param gt_instance_image:

* Normalize input samples' feature. Each sample's feature is normalized via function as follows: * feature[i] = (feature[i] - mean_feature_vector[i])