Go to the documentation of this file.
29 dense_params =
av_malloc(
sizeof(*dense_params));
40 dnn_size += kernel_size * 4;
44 if (dnn_size > file_size || dense_params->
input_num <= 0 ||
51 if (!dense_params->
kernel) {
55 for (
int i = 0;
i < kernel_size; ++
i) {
62 if (!dense_params->
biases){
72 layer->
params = dense_params;
89 int32_t input_operand_index = input_operand_indexes[0];
90 int number = operands[input_operand_index].
dims[0];
91 int height = operands[input_operand_index].
dims[1];
92 int width = operands[input_operand_index].
dims[2];
93 int channel = operands[input_operand_index].
dims[3];
94 const float *
input = operands[input_operand_index].
data;
98 DnnOperand *output_operand = &operands[output_operand_index];
99 output_operand->
dims[0] = number;
105 if (output_operand->
length <= 0) {
110 if (!output_operand->
data) {
118 for (
int y = 0; y <
height; ++y) {
119 for (
int x = 0; x <
width; ++x) {
120 for (
int n_filter = 0; n_filter < dense_params->
output_num; ++n_filter) {
126 for (
int ch = 0; ch < dense_params->
input_num; ++ch) {
128 input_pel =
input[y * src_linesize + x * dense_params->
input_num + ch];
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later. That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another. Frame references ownership and permissions
int ff_dnn_load_layer_dense(Layer *layer, AVIOContext *model_file_context, int file_size, int operands_num)
Load the Densely-Connected Layer.
filter_frame For filters that do not use the this method is called when a frame is pushed to the filter s input It can be called at any time except in a reentrant way If the input frame is enough to produce output
int ff_dnn_execute_layer_dense(DnnOperand *operands, const int32_t *input_operand_indexes, int32_t output_operand_index, const void *parameters, NativeContext *ctx)
Execute the Densely-Connected Layer.
static av_always_inline float av_int2float(uint32_t i)
Reinterpret a 32-bit integer as a float.
int32_t ff_calculate_operand_data_length(const DnnOperand *oprd)
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
void * data
data pointer with data length in bytes.
#define av_assert0(cond)
assert() equivalent, that is always enabled.
DNNDataType data_type
support different kinds of data type such as float, half float, int8 etc, first support float now.
DNNActivationFunc activation
int32_t dims[4]
there are two memory layouts, NHWC or NCHW, so we use dims, dims[0] is Number.
unsigned int avio_rl32(AVIOContext *s)
int32_t output_operand_index
int32_t input_operand_indexes[4]
a layer can have multiple inputs and one output.
and forward the test the status of outputs and forward it to the corresponding return FFERROR_NOT_READY If the filters stores internally one or a few frame for some input
#define i(width, name, range_min, range_max)
void * av_realloc(void *ptr, size_t size)
Allocate, reallocate, or free a block of memory.