aws_detect_faces.py

What is this?

Detect faces within an images using Amazon Rekognition. Please refer to DetectFaces.

Subscribing Topic

• image/compressed (sensor_msgs/CompressedImage)

  Input image. This topic name is resolved from image.

Publishing Topic

• ~faces (opencv_apps/FaceArrayStamped)

  Detected face positions, facial landmarks, emotions, presence of beard, sunglasses, and so on.

• ~poses (geometry_msgs/PoseArray)

  Pose of the face as determined by its pitch, roll, and yaw. The pose is published in Quaternion, to reconstruct Roll, Pitch, Yaw, use following codes.

import math
import rospy
from geometry_msgs.msg import PoseArray

from tf.transformations import euler_from_quaternion

def cb(msg):
    for pose in msg.poses:
        angles = euler_from_quaternion([pose.orientation.x, pose.orientation.y, pose.orientation.z, pose.orientation.w])
        print("roll: {}, pitch: {}, yaw: {}".format(angles[0]*180/math.pi,angles[1]*180/math.pi,angles[2]*180/math.pi))

if __name__ == '__main__':
    rospy.init_node('pose_example')
    rospy.Subscriber('/aws_detect_faces/poses', PoseArray, cb)
    rospy.spin()

• ~attributes (jsk_recognition_msgs/ClassificationResult)

  Attributes of the face, such as emotions, presence of beard, sunglasses, and so on, with confidence.

• ~landmarks (jsk_recognition_msgs/PeoplePoseArray)

  The location of landmarks on the face.

• ~output (sensor_msgs/Image)

  Visualization image of detected faces.

Parameters

• ~aws_credentials_path (String, Default: aws.json)

  Json file for aws access key and secret key to run AWS Rekognition. see Set up an AWS account and create an IAM user for how to obtain keys.

{
    "region" : "xxxxxxxxxxxxxxxxxxxx",
    "aws_access_key_id" : "####################",
    "aws_secret_access_key" : "********************"
}

• ~attributes (Bool, Default: ALL)

  Facial attributes you want to be returned. Set ALL to return all attributes or DEFAULT` to return ``BoundingBox, Confidence, Pose, Quality, and Landmarks.

  see https://github.com/awsdocs/amazon-rekognition-developer-guide/blob/master/doc_source/API_DetectFaces.md#request-parameters more info.

• ~use_window (Bool, Default: False)

  Set true to display a window including detected face and attributes.

• ~buff_size (Int, Default: 614400)

  Buffer size for image subscription. The default buff_size of rospy.Subscriber is 65536, which does not process latest incoming image even if queue_size=1. See https://answers.ros.org/question/220502/image-subscriber-lag-despite-queue-1/ for more detail.

• ~always_subscribe (Bool, Default: True)

  Set true to process even if not one subscribing.

Example

roslaunch jsk_perception sample_aws_detect_faces.launch

Optional Arguments:
  attributes (default "ALL"): set either ALL or DEFAULT for returning attributes
  use_usb_cam (default "false"): set true to use USB camera image as input
  use_window (default "true"): set false if you do not want to display window

For JSK user, Download aws.json file from Gdrive and put this under /tmp directory to run sample code.