Inception-ResNet v1 Face Embedding Operator

author: David Wang

Desription

This operator extracts embedding vector from facial image using Inception-ResNet. The implementation is an adaptation from timesler/facenet-pytorch.

Code Example

Extract face image embedding from './img.png'.

Write the pipeline in simplified style:

import towhee

towhee.glob('./img.png') \
  .image_decode.cv2() \
  .face_embedding.inceptionresnetv1() \
  .tolist()

Write a same pipeline with explicit inputs/outputs name specifications:

import towhee

towhee.glob['path']('./img.png') \
  .image_decode.cv2['path', 'img']() \
  .face_embedding.inceptionresnetv1['img', 'vec']() \
  .select('img','vec') \
  .show()

Factory Constructor

Create the operator via the following factory method

face_embedding.inceptionresnetv1(image_size = 160)

Parameters:

image_size: int

Scaled input image size to extract embedding. The higher resolution would generate the more discriminateive feature but cost more time to calculate.

supported types: int, default is 160.

Interface

A face embedding operator takes a face image as input. It extracts the embedding back to ndarray.

Parameters:

​ img: towhee.types.Image (a sub-class of numpy.ndarray)

​ The input image.

Returns: numpy.ndarray

​ The extracted image embedding.