mirrors/gstreamer
mirrors/gstreamer
1
1
Fork 0
mirror of https://gitlab.freedesktop.org/gstreamer/gstreamer.git synced 2025-02-17 03:35:21 +00:00
Code Issues 1 Projects Releases Wiki Activity

onnx: Add more tensor data types

Browse source 
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/6001>
This commit is contained in:
Olivier Crête 2024-01-24 21:31:44 -05:00
parent e3d8168a5a
commit 13de5160be
3 changed files with 35 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
subprojects/gst-plugins-bad/ext/onnx/gstonnxclient.cpp
View file

@ -53,7 +53,7 @@ GstOnnxClient::GstOnnxClient (GstElement *debug_parent):debug_parent(debug_paren
dest (nullptr),
m_provider (GST_ONNX_EXECUTION_PROVIDER_CPU),
inputImageFormat (GST_ML_INPUT_IMAGE_FORMAT_HWC),
inputDatatype (GST_TENSOR_TYPE_INT8),
inputDatatype (GST_TENSOR_TYPE_UINT8),
inputDatatypeSize (sizeof (uint8_t)),
fixedInputImageSize (false),
inputTensorOffset (0.0),
@ -100,21 +100,27 @@ GstOnnxClient::GstOnnxClient (GstElement *debug_parent):debug_parent(debug_paren
{
inputDatatype = datatype;
switch (inputDatatype) {
case GST_TENSOR_TYPE_INT8:
case GST_TENSOR_TYPE_UINT8:
inputDatatypeSize = sizeof (uint8_t);
break;
case GST_TENSOR_TYPE_INT16:
case GST_TENSOR_TYPE_UINT16:
inputDatatypeSize = sizeof (uint16_t);
break;
case GST_TENSOR_TYPE_INT32:
case GST_TENSOR_TYPE_UINT32:
inputDatatypeSize = sizeof (uint32_t);
break;
case GST_TENSOR_TYPE_INT32:
inputDatatypeSize = sizeof (int32_t);
break;
case GST_TENSOR_TYPE_FLOAT16:
inputDatatypeSize = 2;
break;
case GST_TENSOR_TYPE_FLOAT32:
inputDatatypeSize = sizeof (float);
break;
default:
g_error ("Data type %d not handled", inputDatatype);
break;
};
}
@ -241,7 +247,7 @@ GstOnnxClient::GstOnnxClient (GstElement *debug_parent):debug_parent(debug_paren
switch (elementType) {
case ONNX_TENSOR_ELEMENT_DATA_TYPE_UINT8:
setInputImageDatatype(GST_TENSOR_TYPE_INT8);
setInputImageDatatype(GST_TENSOR_TYPE_UINT8);
break;
case ONNX_TENSOR_ELEMENT_DATA_TYPE_FLOAT:
setInputImageDatatype(GST_TENSOR_TYPE_FLOAT32);
@ -450,7 +456,7 @@ GstOnnxClient::GstOnnxClient (GstElement *debug_parent):debug_parent(debug_paren
std::vector < Ort::Value > inputTensors;
switch (inputDatatype) {
case GST_TENSOR_TYPE_INT8:
case GST_TENSOR_TYPE_UINT8:
uint8_t *src_data;
if (inputTensorOffset == 00 && inputTensorScale == 1.0) {
src_data = img_data;

2
subprojects/gst-plugins-bad/ext/onnx/gstonnxinference.cpp
View file

@ -489,7 +489,7 @@ gst_onnx_inference_transform_caps (GstBaseTransform *
onnxClient->getWidth (), "height", G_TYPE_INT,
onnxClient->getHeight (), NULL);
if (onnxClient->getInputImageDatatype() == GST_TENSOR_TYPE_INT8 &&
if (onnxClient->getInputImageDatatype() == GST_TENSOR_TYPE_UINT8 &&
onnxClient->getInputImageScale() == 1.0 &&
onnxClient->getInputImageOffset() == 0.0) {
switch (onnxClient->getChannels()) {

26
subprojects/gst-plugins-bad/ext/onnx/tensor/gsttensormeta.h
View file

@ -27,21 +27,39 @@
/**
* GstTensorType:
*
* @GST_TENSOR_TYPE_INT8 8 bit integer tensor data
* @GST_TENSOR_TYPE_INT16 16 bit integer tensor data
* @GST_TENSOR_TYPE_INT32 32 bit integer tensor data
* @GST_TENSOR_TYPE_INT4 signed 4 bit integer tensor data
* @GST_TENSOR_TYPE_INT8 signed 8 bit integer tensor data
* @GST_TENSOR_TYPE_INT16 signed 16 bit integer tensor data
* @GST_TENSOR_TYPE_INT32 signed 32 bit integer tensor data
* @GST_TENSOR_TYPE_INT64 signed 64 bit integer tensor data
* @GST_TENSOR_TYPE_UINT4 unsigned 4 bit integer tensor data
* @GST_TENSOR_TYPE_UINT8 unsigned 8 bit integer tensor data
* @GST_TENSOR_TYPE_UINT16 unsigned 16 bit integer tensor data
* @GST_TENSOR_TYPE_UINT32 unsigned 32 bit integer tensor data
* @GST_TENSOR_TYPE_UINT64 unsigned 64 bit integer tensor data
* @GST_TENSOR_TYPE_FLOAT16 16 bit floating point tensor data
* @GST_TENSOR_TYPE_FLOAT32 32 bit floating point tensor data
* @GST_TENSOR_TYPE_FLOAT64 64 bit floating point tensor data
* @GST_TENSOR_TYPE_BFLOAT16 "brain" 16 bit floating point tensor data
*
* Since: 1.24
*/
typedef enum _GstTensorType
{
GST_TENSOR_TYPE_INT4,
GST_TENSOR_TYPE_INT8,
GST_TENSOR_TYPE_INT16,
GST_TENSOR_TYPE_INT32,
GST_TENSOR_TYPE_INT64,
GST_TENSOR_TYPE_UINT4,
GST_TENSOR_TYPE_UINT8,
GST_TENSOR_TYPE_UINT16,
GST_TENSOR_TYPE_UINT32,
GST_TENSOR_TYPE_UINT64,
GST_TENSOR_TYPE_FLOAT16,
GST_TENSOR_TYPE_FLOAT32
GST_TENSOR_TYPE_FLOAT32,
GST_TENSOR_TYPE_FLOAT64,
GST_TENSOR_TYPE_BFLOAT16,
} GstTensorType;