Postprecess the prediction result. Decode detection result, set the confidence threshold and do the NMS to keep the appropriate detect

Process the predicted landmarks into the form of the original image. Returns: A numpy array, the landmarks based on the shape of

Preprocess the image, such as standardization and other operations. Returns: A numpy array list, the shape is channel * h * w.

Preprocess the image, such as standardization and other operations. Returns: A tensor, the shape is 1 x 3 x h x w. A

Preprocess the input image, cutting the input image through the face detection information. Using the face detection result(dets) to get the f

Preprocess the input image. Returns: A torch tensor, the input after preprecess, shape: (3, 112, 112).

make a grid image from an image batch. Args: images (torch.Tensor): input image batch. nrows: rows of grid. border: borde

Build optimizer, set weight decay of normalization to 0 by default.

Decode a list of string notations to specify blocks inside the network. Args: string_list (list[str]): A list of strings, each st

Decode locations from predictions using priors to undo the encoding we did for offset regression at train time. Args: loc

Drop connect. Args: input (tensor: BCWH): Input of this structure. p (float: 0.0~1.0): Probability of drop connection. tr

Extract face + margin from PIL Image given bounding box. Arguments: img {PIL.Image} -- A PIL Image. box {numpy.ndarray} -- Fo

use convolution layer to extract feature . Args: inputs (tensor): Input tensor. Returns: Output of the final

Calculate total NME for a batch of data Args: pred_heatmaps: torch tensor of size [batch, points, height, width]

create an efficientnet model according to name. Args: model_name (str): Name for efficientnet. in_channels (int): Inp

Generate the train file, which has the following format. relative_path0 label0 relative_path1 label1 relative_path2 label2

Get the best threshold by train_score_list and train_label_list. Args: score_list(ndarray): the score list of all pairs.

Get the best threshold by train_score_list and train_label_list. Args: score_list(ndarray): the score list of all pairs.

Args: vertices: 3 x nver triangles: 3 x ntri h: height w: width Returns: depth_buffer: height x width

Get the feature list of facescrub. Returns: A numpy array, the shape is N * feat_dim, N is the number of images in facescrub.

Get the feature list of facescrub. Returns: A numpy array, the shape is N * feat_dim, N is the number of images in facescrub.

Get the feature list of each id in facescrub. Returns: A dict，key is the id in facescrub，value is the feature list of this id.

Get the feature list of each id in facescrub. Returns: A dict，key is the id in facescrub，value is the feature list of this id.

Generate to do task list. Args: face_info_file: The file which contains image_name and landmarks. image_name2template_name_file:

Get the current learning rate from optimizer.

Get the current learning rate from optimizer.

Get the current learning rate from optimizer.

Get the current learning rate from optimizer.