Skip to content

Latest commit

 

History

History
100 lines (87 loc) · 3.01 KB

README.md

File metadata and controls

100 lines (87 loc) · 3.01 KB

GoogleNet

Model Download Download (with sample test data) ONNX version Opset version
GoogleNet 28 MB 31 MB 1.1 3
GoogleNet 28 MB 31 MB 1.1.2 6
GoogleNet 28 MB 31 MB 1.2 7
GoogleNet 28 MB 31 MB 1.3 8
GoogleNet 28 MB 31 MB 1.4 9

Description

GoogLeNet is the name of a convolutional neural network for classification, which competed in the ImageNet Large Scale Visual Recognition Challenge in 2014.

Differences:

  • not training with the relighting data-augmentation;
  • not training with the scale or aspect-ratio data-augmentation;
  • uses "xavier" to initialize the weights instead of "gaussian";

Paper

Going deeper with convolutions

Dataset

ILSVRC2014

Source

Caffe BVLC GoogLeNet ==> Caffe2 GoogLeNet ==> ONNX GoogLeNet

Model input and output

Input

data_0: float[1, 3, 224, 224]

Output

prob_0: float[1, 1000]

Pre-processing steps

Necessary Imports

import imageio
from PIL import Image

Obtain and pre-process image

def get_image(path):
'''
    Using path to image, return the RGB load image
'''
    img = imageio.imread(path, pilmode='RGB')
    return img
    
# Pre-processing function for ImageNet models using numpy
def preprocess(img):   
    '''
    Preprocessing required on the images for inference with mxnet gluon
    The function takes loaded image and returns processed tensor
    '''
    img = np.array(Image.fromarray(img).resize((224, 224))).astype(np.float32)
    img[:, :, 0] -= 123.68
    img[:, :, 1] -= 116.779
    img[:, :, 2] -= 103.939
    img[:,:,[0,1,2]] = img[:,:,[2,1,0]]
    img = img.transpose((2, 0, 1))
    img = np.expand_dims(img, axis=0)
    
    return img

Post-processing steps

def predict(path):
    # based on : https://mxnet.apache.org/versions/1.0.0/tutorials/python/predict_image.html
    img = get_image(path)
    img = preprocess(img)
    mod.forward(Batch([mx.nd.array(img)]))
    # Take softmax to generate probabilities
    prob = mod.get_outputs()[0].asnumpy()
    prob = np.squeeze(prob)
    a = np.argsort(prob)[::-1]
    return a

Sample test data

random generated sample test data:

  • test_data_set_0
  • test_data_set_1
  • test_data_set_2
  • test_data_set_3
  • test_data_set_4
  • test_data_set_5

Results/accuracy on test set

This bundled model obtains a top-1 accuracy 68.7% (31.3% error) and a top-5 accuracy 88.9% (11.1% error) on the validation set, using just the center crop. (Using the average of 10 crops, (4 + 1 center) * 2 mirror, should obtain a bit higher accuracy.)

License

BSD-3