BEAT - anjos/crop

This algorithm is a legacy one. The API has changed since its implementation. New versions and forks will need to be updated.

This algorithm is splittable

Endpoint Groups 1

Algorithms have at least one input and one output. All algorithm endpoints are organized in groups. Groups are used by the platform to indicate which inputs and outputs are synchronized together. The first group is automatically synchronized with the channel defined by the block in which the algorithm is deployed.

Group: main

Endpoint Name	Data Format	Nature
video	system/array_4d_uint8/1	Input
annotations	system/bounding_box_video/1	Input
cropped_faces	system/array_3d_uint8/1	Output

Parameters 3

Parameters allow users to change the configuration of an algorithm when scheduling an experiment

Name	Description	Type	Default
normed-height	Height of the face bounding box after size normalization	uint32	64
normed-width	Width of the face bounding box after size normalization	uint32	64
minimum-face-size	Frames with face bounding box smaller than this will be discarded	uint32	50

import numpy
import bob

class Algorithm:

def __init__(self):
        self.normed_height = 64
        self.normed_width  = 64
        self.minimum_face_size = 50 # minimum size of a valid face

def setup(self, parameters):
        self.normed_height = parameters.get('crop-height', self.normed_height)
        self.normed_width  = parameters.get('crop-width',  self.normed_width)
        self.minimum_face_size = parameters.get('minimum-face-size', self.minimum_face_size)

return True

def process(self, inputs, outputs):

video = inputs["video"].data.value
        annotations = inputs["annotations"].data.value

output = []
 
        for i in range(video.shape[0]): #range(video.shape[0], -1, -1):
          frame = bob.ip.rgb_to_gray(video[i,:,:,:])

if annotations[i].height < self.minimum_face_size or annotations[i].width < self.minimum_face_size:
            continue # don't add the invalid frames

cropped_face = bob.ip.crop(frame,
                                     int(getattr(annotations[i], 'top-left-y')), 
                                     int(getattr(annotations[i], 'top-left-x')), 
                                     int(annotations[i].height), 
                                     int(annotations[i].width),
                                    )

tempbbx = numpy.zeros((self.normed_height, self.normed_width), dtype='float64')
          bob.ip.scale(cropped_face, tempbbx) # normalization
          tempbbx += 0.5
          output.append(numpy.floor(tempbbx).astype('uint8'))

outputs["cropped_faces"].write({'value': numpy.array(output)})

return True

The code for this algorithm in Python
The ruler at 80 columns indicate suggested POSIX line breaks (for readability).
The editor will automatically enlarge to accomodate the entirety of your input
Use keyboard shortcuts for search/replace and faster editing. For example, use Ctrl-F (PC) or Cmd-F (Mac) to search through this box

This algorithm performs pre-processing of the input videos in two steps:

Cropping the face bounding box from each frame of the video based on the input face files. Faces with face bounding box smaller than minimum-face-size parameter will be discarded.
Normalizes the face bouding boxes by size, depending on the parameters normed-width and normed-height.

This algorithm relies on the Bob library for cropping and scaling the video frame images.

Experiments

Attestation:

Privacy:

Status:

			Updated	Name	Databases/Protocols	Analyzers
				anjos/ivana7c/simple-antispoofing-updated/1/face-antipoofing-lbp-histogram-comparison	replay/1@countermeasure	anjos/antispoofing_analyzer/1

Scientific Python 2.7 (0.0.4) 15

This table shows the number of times this algorithm has been successfully run using the given environment. Note this does not provide sufficient information to evaluate if the algorithm will run when submitted to different conditions.

algorithms anjos crop_face 1

Endpoint Groups 1

Group: main

Parameters 3

Experiments

algorithms

anjos

crop_face

1