| | |
| | | l.w = w; |
| | | l.c = c; |
| | | l.pad = padding; |
| | | l.out_w = (w + 2*padding)/stride; |
| | | l.out_h = (h + 2*padding)/stride; |
| | | l.out_w = (w + padding - size) / stride + 1; |
| | | l.out_h = (h + padding - size) / stride + 1; |
| | | l.out_c = c; |
| | | l.outputs = l.out_h * l.out_w * l.out_c; |
| | | l.inputs = h*w*c; |
| | |
| | | l.output_gpu = cuda_make_array(l.output, output_size); |
| | | l.delta_gpu = cuda_make_array(l.delta, output_size); |
| | | #endif |
| | | fprintf(stderr, "max %d x %d / %d %4d x%4d x%4d -> %4d x%4d x%4d\n", size, size, stride, w, h, c, l.out_w, l.out_h, l.out_c); |
| | | l.bflops = (l.size*l.size*l.c * l.out_h*l.out_w) / 1000000000.; |
| | | fprintf(stderr, "max %d x %d / %d %4d x%4d x%4d -> %4d x%4d x%4d %5.3f BF\n", size, size, stride, w, h, c, l.out_w, l.out_h, l.out_c, l.bflops); |
| | | return l; |
| | | } |
| | | |
| | |
| | | l->w = w; |
| | | l->inputs = h*w*l->c; |
| | | |
| | | l->out_w = (w + 2*l->pad)/l->stride; |
| | | l->out_h = (h + 2*l->pad)/l->stride; |
| | | l->out_w = (w + l->pad - l->size) / l->stride + 1; |
| | | l->out_h = (h + l->pad - l->size) / l->stride + 1; |
| | | l->outputs = l->out_w * l->out_h * l->c; |
| | | int output_size = l->outputs * l->batch; |
| | | |
| | |
| | | void forward_maxpool_layer(const maxpool_layer l, network_state state) |
| | | { |
| | | int b,i,j,k,m,n; |
| | | int w_offset = -l.pad; |
| | | int h_offset = -l.pad; |
| | | int w_offset = -l.pad / 2; |
| | | int h_offset = -l.pad / 2; |
| | | |
| | | int h = l.out_h; |
| | | int w = l.out_w; |