| | |
| | | return float_to_image(w,h,c,l.delta); |
| | | } |
| | | |
| | | convolutional_layer make_convolutional_layer(int batch, int h, int w, int c, int n, int size, int stride, int pad, ACTIVATION activation) |
| | | convolutional_layer make_convolutional_layer(int batch, int h, int w, int c, int n, int size, int stride, int pad, ACTIVATION activation, int batch_normalize) |
| | | { |
| | | int i; |
| | | convolutional_layer l = {0}; |
| | |
| | | l.stride = stride; |
| | | l.size = size; |
| | | l.pad = pad; |
| | | l.batch_normalize = batch_normalize; |
| | | |
| | | l.filters = calloc(c*n*size*size, sizeof(float)); |
| | | l.filter_updates = calloc(c*n*size*size, sizeof(float)); |
| | | |
| | | l.biases = calloc(n, sizeof(float)); |
| | | l.bias_updates = calloc(n, sizeof(float)); |
| | | |
| | | // float scale = 1./sqrt(size*size*c); |
| | | float scale = sqrt(2./(size*size*c)); |
| | | for(i = 0; i < c*n*size*size; ++i) l.filters[i] = 2*scale*rand_uniform() - scale; |
| | | for(i = 0; i < n; ++i){ |
| | | l.biases[i] = scale; |
| | | } |
| | | int out_h = convolutional_out_height(l); |
| | | int out_w = convolutional_out_width(l); |
| | | l.out_h = out_h; |
| | |
| | | l.output = calloc(l.batch*out_h * out_w * n, sizeof(float)); |
| | | l.delta = calloc(l.batch*out_h * out_w * n, sizeof(float)); |
| | | |
| | | if(batch_normalize){ |
| | | l.scales = calloc(n, sizeof(float)); |
| | | l.scale_updates = calloc(n, sizeof(float)); |
| | | for(i = 0; i < n; ++i){ |
| | | l.scales[i] = 1; |
| | | } |
| | | |
| | | l.mean = calloc(n, sizeof(float)); |
| | | l.spatial_mean = calloc(n*l.batch, sizeof(float)); |
| | | |
| | | l.variance = calloc(n, sizeof(float)); |
| | | l.rolling_mean = calloc(n, sizeof(float)); |
| | | l.rolling_variance = calloc(n, sizeof(float)); |
| | | } |
| | | |
| | | #ifdef GPU |
| | | l.filters_gpu = cuda_make_array(l.filters, c*n*size*size); |
| | | l.filter_updates_gpu = cuda_make_array(l.filter_updates, c*n*size*size); |
| | |
| | | l.biases_gpu = cuda_make_array(l.biases, n); |
| | | l.bias_updates_gpu = cuda_make_array(l.bias_updates, n); |
| | | |
| | | l.scales_gpu = cuda_make_array(l.scales, n); |
| | | l.scale_updates_gpu = cuda_make_array(l.scale_updates, n); |
| | | |
| | | l.col_image_gpu = cuda_make_array(l.col_image, out_h*out_w*size*size*c); |
| | | l.delta_gpu = cuda_make_array(l.delta, l.batch*out_h*out_w*n); |
| | | l.output_gpu = cuda_make_array(l.output, l.batch*out_h*out_w*n); |
| | | |
| | | if(batch_normalize){ |
| | | l.mean_gpu = cuda_make_array(l.mean, n); |
| | | l.variance_gpu = cuda_make_array(l.variance, n); |
| | | |
| | | l.rolling_mean_gpu = cuda_make_array(l.mean, n); |
| | | l.rolling_variance_gpu = cuda_make_array(l.variance, n); |
| | | |
| | | l.spatial_mean_gpu = cuda_make_array(l.spatial_mean, n*l.batch); |
| | | l.spatial_variance_gpu = cuda_make_array(l.spatial_mean, n*l.batch); |
| | | |
| | | l.spatial_mean_delta_gpu = cuda_make_array(l.spatial_mean, n*l.batch); |
| | | l.spatial_variance_delta_gpu = cuda_make_array(l.spatial_mean, n*l.batch); |
| | | |
| | | l.mean_delta_gpu = cuda_make_array(l.mean, n); |
| | | l.variance_delta_gpu = cuda_make_array(l.variance, n); |
| | | |
| | | l.x_gpu = cuda_make_array(l.output, l.batch*out_h*out_w*n); |
| | | l.x_norm_gpu = cuda_make_array(l.output, l.batch*out_h*out_w*n); |
| | | } |
| | | #endif |
| | | l.activation = activation; |
| | | |
| | |
| | | return l; |
| | | } |
| | | |
| | | void denormalize_convolutional_layer(convolutional_layer l) |
| | | { |
| | | int i, j; |
| | | for(i = 0; i < l.n; ++i){ |
| | | float scale = l.scales[i]/sqrt(l.rolling_variance[i] + .00001); |
| | | for(j = 0; j < l.c*l.size*l.size; ++j){ |
| | | l.filters[i*l.c*l.size*l.size + j] *= scale; |
| | | } |
| | | l.biases[i] -= l.rolling_mean[i] * scale; |
| | | } |
| | | } |
| | | |
| | | void test_convolutional_layer() |
| | | { |
| | | convolutional_layer l = make_convolutional_layer(1, 5, 5, 3, 2, 5, 2, 1, LEAKY, 1); |
| | | l.batch_normalize = 1; |
| | | float data[] = {1,1,1,1,1, |
| | | 1,1,1,1,1, |
| | | 1,1,1,1,1, |
| | | 1,1,1,1,1, |
| | | 1,1,1,1,1, |
| | | 2,2,2,2,2, |
| | | 2,2,2,2,2, |
| | | 2,2,2,2,2, |
| | | 2,2,2,2,2, |
| | | 2,2,2,2,2, |
| | | 3,3,3,3,3, |
| | | 3,3,3,3,3, |
| | | 3,3,3,3,3, |
| | | 3,3,3,3,3, |
| | | 3,3,3,3,3}; |
| | | network_state state = {0}; |
| | | state.input = data; |
| | | forward_convolutional_layer(l, state); |
| | | } |
| | | |
| | | void resize_convolutional_layer(convolutional_layer *l, int w, int h) |
| | | { |
| | | l->w = w; |
| | |
| | | } |
| | | } |
| | | |
| | | |
| | | void forward_convolutional_layer(const convolutional_layer l, network_state state) |
| | | { |
| | | int out_h = convolutional_out_height(l); |
| | |
| | | c += n*m; |
| | | state.input += l.c*l.h*l.w; |
| | | } |
| | | |
| | | if(l.batch_normalize){ |
| | | mean_cpu(l.output, l.batch, l.n, l.out_h*l.out_w, l.mean); |
| | | variance_cpu(l.output, l.mean, l.batch, l.n, l.out_h*l.out_w, l.variance); |
| | | normalize_cpu(l.output, l.mean, l.variance, l.batch, l.n, l.out_h*l.out_w); |
| | | } |
| | | |
| | | activate_array(l.output, m*n*l.batch, l.activation); |
| | | } |
| | | |