Ref:https://github.com/1adrianb/face-alignment/blob/master/face_alignment/utils.py

3x3 convolution with padding

Numpy array to tensor. Args: imgs (list[ndarray] | ndarray): Input images. bgr2rgb (bool): Whether to change bgr to rgb.

Function: ---------- apply affine transform 'trans' to uv Parameters: ---------- @trans: 3x3 np.array tr

This method must be called once for each frame even with empty detections. NOTE:as in practical realtime MOT, the detector doesn't run

Align and warp faces with face template.

3x3 convolution with padding

Takes a bounding box in the form [x1,y1,x2,y2] and returns z in the form [x,y,s,r] where x,y is the centre of the box and s is the scale/area and

Takes a bounding box in the centre form [x,y,s,r] and returns it in the form [x1,y1,x2,y2] where x1,y1 is the top left and x2,y2 is the bottom

Function: ---------- get reference 5 key points according to crop settings: 0. Set default crop_size: if default_

Function: ---------- Find Similarity Transform Matrix 'trans': u = src_pts[:, 0] v = src_pts[:, 1]

Write image to file. Args: img (ndarray): Image array to be written. file_path (str): Image file path. params (None or li

Pure Python NMS baseline.

img can be image path or cv2 loaded image.

Updates the state vector with observed bbox.

Visualize detection results. Args: img (Numpy array): Input image. CHW, BGR, [0, 255], uint8.

Align and crop face with landmarks. The output_size and transform_size are based on width. The height is adjusted based on shrink_ratio_h/shr

Assigns detections to tracked object (both represented as bounding boxes) Returns: 3 lists of matches, unmatched_detections and unmatched

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

Decode landm from predictions using priors to undo the encoding we did for offset regression at train time. Args: pre (tensor): landm

Arguments: frames: a list of PIL.Image, or torch.Tensor(shape=[n, h, w, c], type=np.float32, BGR format).

Function: ---------- Convert Transform Matrix 'trans' into 'cv2_trans' which could be directly used by cv2.warpAffine():

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

Decode landm from predictions using priors to undo the encoding we did for offset regression at train time. Args: pre (tensor): landm

Visualize results

draw head pose axis.

draw head pose cube. Where (tdx, tdy) is the translation of the face. For pose we have [pitch yaw roll tdx tdy tdz scale_factor]

Encode the variances from the priorbox layers into the ground truth boxes we have matched (based on jaccard overlap) with the prior boxes. Arg

Encode the variances from the priorbox layers into the ground truth boxes we have matched (based on jaccard overlap) with the prior boxes. Arg

Function: ---------- get affine transform matrix 'tfm' from src_pts to dst_pts Parameters: ---------- @src_pts: Kx2 n

Function: ---------- Find Similarity Transform Matrix 'cv2_trans' which could be directly used by cv2.warpAffine():

Returns the current bounding box estimate.

We resize both tensors to [A,B,2] without new malloc: [A,2] -> [A,1,2] -> [A,B,2] [B,2] -> [1,B,2] -> [A,B,2] Then we compute the area of

Computes IOU between two bboxes in the form [x1,y1,x2,y2]

Compute the jaccard overlap of two sets of boxes. The jaccard overlap is simply the intersection over union of two boxes. Here we operate on

Transfer 98 landmark positions to 68 landmark positions. Args: landmark_98(numpy array): Polar coordinates of 98 landmarks, (98, 2) Re

Load data or string from txt file.