Can you get landmarks from native post-processing?
2.If the answer to the above question is yes, do you want to get the corresponding landmark in Python?
This is a sample for facial landmark.
NvDsUserMeta *user_meta = (NvDsUserMeta *) l_user->data;
if (user_meta->base_meta.meta_type != NVDSINFER_TENSOR_OUTPUT_META)
continue;
NvDsInferTensorMeta *meta =
(NvDsInferTensorMeta *) user_meta->user_meta_data;
float * heatmap_data = NULL;
float * confidence = NULL;
//int heatmap_c = 0;
for (unsigned int i = 0; i < meta->num_output_layers; i++) {
NvDsInferLayerInfo *info = &meta->output_layers_info[i];
info->buffer = meta->out_buf_ptrs_host[i];
std::vector < NvDsInferLayerInfo >
outputLayersInfo (meta->output_layers_info,
meta->output_layers_info + meta->num_output_layers);
//Prepare CVCORE input layers
if (strcmp(outputLayersInfo[i].layerName,
"softargmax") == 0) {
//This layer output landmarks coordinates
This is another sample for add meta python bindings.
35.How to add custom metadata in a GStreamer native plugin and access it in Python.
Here is an example based on deepstream-test1.py and dsexample.
35.1. Add custom metadata in the dsexample plugin.
diff --git a/sources/gst-plugins/gst-dsexample/gstdsexample.cpp b/sources/gst-plugins/gst-dsexample/gstdsexample.cpp
index d5399c7..764ec0a 100644
--- a/sources/gst-plugins/gst-dsexample/gstdsexample.cpp
+++ b/sources/gst-plugins/gst-dsexample/gstdsexample.cpp
@@ -318,6 +318,76 @@ gst_dsexample_get…