• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  * Copyright (c) 2018 Sergey Lavrushkin
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * FFmpeg is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
14  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with FFmpeg; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 /**
22  * @file
23  * DNN native backend implementation.
24  */
25 
26 #include "dnn_backend_native.h"
27 #include "dnn_backend_native_layer_depth2space.h"
28 
ff_dnn_load_layer_depth2space(Layer * layer,AVIOContext * model_file_context,int file_size,int operands_num)29 int ff_dnn_load_layer_depth2space(Layer *layer, AVIOContext *model_file_context, int file_size, int operands_num)
30 {
31     DepthToSpaceParams *params;
32     int dnn_size = 0;
33     params = av_malloc(sizeof(*params));
34     if (!params)
35         return 0;
36 
37     params->block_size = (int32_t)avio_rl32(model_file_context);
38     dnn_size += 4;
39     layer->input_operand_indexes[0] = (int32_t)avio_rl32(model_file_context);
40     layer->output_operand_index = (int32_t)avio_rl32(model_file_context);
41     dnn_size += 8;
42     layer->params = params;
43 
44     if (layer->input_operand_indexes[0] >= operands_num || layer->output_operand_index >= operands_num) {
45         return 0;
46     }
47 
48     return dnn_size;
49 }
50 
ff_dnn_execute_layer_depth2space(DnnOperand * operands,const int32_t * input_operand_indexes,int32_t output_operand_index,const void * parameters,NativeContext * ctx)51 int ff_dnn_execute_layer_depth2space(DnnOperand *operands, const int32_t *input_operand_indexes,
52                                      int32_t output_operand_index, const void *parameters, NativeContext *ctx)
53 {
54     float *output;
55     const DepthToSpaceParams *params = parameters;
56     int block_size = params->block_size;
57     int32_t input_operand_index = input_operand_indexes[0];
58     int number = operands[input_operand_index].dims[0];
59     int height = operands[input_operand_index].dims[1];
60     int width = operands[input_operand_index].dims[2];
61     int channels = operands[input_operand_index].dims[3];
62     const float *input = operands[input_operand_index].data;
63 
64     int y, x, by, bx, ch;
65     int new_channels = channels / (block_size * block_size);
66     int output_linesize = width * channels;
67     int by_linesize = output_linesize / block_size;
68     int x_linesize = new_channels * block_size;
69 
70     DnnOperand *output_operand = &operands[output_operand_index];
71     output_operand->dims[0] = number;
72     output_operand->dims[1] = height * block_size;
73     output_operand->dims[2] = width * block_size;
74     output_operand->dims[3] = new_channels;
75     output_operand->data_type = operands[input_operand_index].data_type;
76     output_operand->length = ff_calculate_operand_data_length(output_operand);
77     if (output_operand->length <= 0) {
78         av_log(ctx, AV_LOG_ERROR, "The output data length overflow\n");
79         return AVERROR(EINVAL);
80     }
81     output_operand->data = av_realloc(output_operand->data, output_operand->length);
82     if (!output_operand->data) {
83         av_log(ctx, AV_LOG_ERROR, "Failed to reallocate memory for output\n");
84         return AVERROR(ENOMEM);
85     }
86     output = output_operand->data;
87 
88     for (y = 0; y < height; ++y){
89         for (x = 0; x < width; ++x){
90             for (by = 0; by < block_size; ++by){
91                 for (bx = 0; bx < block_size; ++bx){
92                     for (ch = 0; ch < new_channels; ++ch){
93                         output[by * by_linesize + x * x_linesize + bx * new_channels + ch] = input[ch];
94                     }
95                     input += new_channels;
96                 }
97             }
98         }
99         output += output_linesize;
100     }
101     return 0;
102 }
103