Repository files navigation

For Course CS205 'C/C++ Program Design' at Southern University of Scicence and Technology, China.

The model is trained to perform face classification (face or background).

Detailed definition:model.py. Visualization:netron (NOTE: you need an extra softmax layer in the end of the pipepline to output scores in the range [0.0, 1.0]).

More aboutface_binary_cls.cpp:

• This file is ported fromface_binary_cls.pth usingport2cpp defined inmodel.py.
• Input: a tensor,
  • loaded from an 128x128 RGB image asRGB format andshape [channel, height, width],
  • normalized in the range [0.0, 1.0].
• Output: a tensor ofshape [2]. Softmax is needed to compute confidences in the range [0.0, 1.0]. Values at index 0 stands for the confidence of background, while index 1 for face's.
• Note that the parameters of batch normalization is already combined to convolutional layers' when porting weights (.pth) to.cpp.

A convolutional layer (conv) is defined as[out_channels, in_channels, kernel_size_h, kernel_size_w]. It takes a tensor of shape[in_channels, in_h, in_w] as input, and ouputs a tensor of shape[out_channels, out_h, out_w]. Example of locating weights and bias for a 3x3 kernel atout_channels=o, in_channels=i:

for (int o =0; o < out_channels; ++o) {for (int i =0; i < in_channels; ++i) {// weights// first row of the kernelfloat kernel_oi_00 = conv0_weight[o*(in_channels*3*3) + i*(3*3) +0];float kernel_oi_01 = conv0_weight[o*(in_channels*3*3) + i*(3*3) +1];float kernel_oi_02 = conv0_weight[o*(in_channels*3*3) + i*(3*3) +2];// and more rows ...// biasfloat bias_oi = conv0_bias[o];    }}

A fully connected layer (fc) is defined as[out_features, in_features]. It takes a tensor of shape[N, in_features] as input, and outputs a tensor of shape[N, out_features].N is denoted as batch size, batch size is 1 if there is one image in the input. The calculation of the fully connected layer is matrix multiplication.For the weight matrix of shape[out_features, in_features], you can iterate as follows:

for (int o =0; o < out_features; ++o) {for (int i =0; i < in_features; ++i) {float w_oi = fc0_weight[o*out_features + i];// ...    }float bias = fc0_bias[o];}

We provide a demo to output scores as an example indemo.py using PyTorch (>= 1.6.0) and two sample images insamples. You can run the demo and get the confidence scores as follows:

$ python demo.py --img ./samples/face.jpgbg score: 0.007086, face score: 0.992914.$ python demo.py --img ./samples/bg.jpgbg score: 0.999996, face score: 0.000004.

ThankYuantao Feng to train the model.