Please provide complete information as applicable to your setup.
• Hardware Platform (Jetson / GPU) :- dGPU
• DeepStream Version :- 6.4
I have a pipeline that looks like this
streammux -> nvvideoconv2 -> caps_filter -> darknet_infer -> nvvidconv -> caps_filter2 ->nvosd -> queue6 -> sink
Darknet infer is the gst-darknet model link.
I refered to deepstream infer meta test sample.
What I want to do is, I want to take the objects detected by gst-darknet and convert them to NvDsObjectMeta.
I have attached following probe function at nvvidconv sink pad.
static GstPadProbeReturn converter_probe(GstPad *pad,
GstPadProbeInfo *info,
gpointer u_data) {
GstBuffer *buf = (GstBuffer *)info->data;
NvDsBatchMeta *batch_meta =
gst_buffer_get_nvds_batch_meta(GST_BUFFER(info->data));
// nvds_acquire_meta_lock(batch_meta);
g_print("[Attacher Probe]Batch meta : num frames %d \n",
batch_meta->num_frames_in_batch);
static const GstMetaInfo *metainfo = NULL;
if (metainfo == NULL) {
metainfo = gst_meta_get_info("GstDarknetMetaDetections");
}
GstDarknetMetaDetections *meta =
(GstDarknetMetaDetections *)gst_buffer_get_meta(buf, metainfo->api);
// detection count is available in the detection_count variable
// g_print("detections: %d\n", meta->detection_count);
for (NvDsMetaList *l_frame = batch_meta->frame_meta_list; l_frame != NULL;
l_frame = l_frame->next) {
NvDsFrameMeta *frame_meta = (NvDsFrameMeta *)l_frame->data;
// detections are available in the detections variable
for (guint i = 0; i < meta->detection_count; i++) {
GstDarknetMetaDetection *det = &meta->detections[i];
NvDsObjectMeta *obj_meta =
nvds_acquire_obj_meta_from_pool(batch_meta);
obj_meta->unique_component_id =
0; // TODO : Figure out a better way to fill this field
obj_meta->confidence = det->probability;
obj_meta->object_id =
UNTRACKED_OBJECT_ID;
obj_meta->class_id = det->classid;
NvOSD_RectParams &rect_params = obj_meta->rect_params;
NvOSD_TextParams &text_params = obj_meta->text_params;
/* Assign bounding box coordinates. These can be overwritten if
* tracker component is present in the pipeline */
g_print("[Darknet Detection ] [%d %d %d %d ] \n", det->xmin,
det->ymin, det->xmax, det->ymax);
rect_params.left = det->xmin;
rect_params.top = det->ymin;
rect_params.width = det->xmax - det->xmin;
rect_params.height = det->ymax - det->ymin;
/* Border of width 3. */
rect_params.border_width = 3;
rect_params.has_bg_color = 0;
rect_params.border_color = (NvOSD_ColorParams){1, 0, 0, 1};
/* display_text requires heap allocated memory. */
// text_params.display_text;
/* Display text above the left top corner of the object. */
text_params.x_offset = rect_params.left;
text_params.y_offset = rect_params.top - 10;
/* Set black background for the text. */
text_params.set_bg_clr = 1;
text_params.text_bg_clr = (NvOSD_ColorParams){0, 0, 0, 1};
/* Font face, size and color. */
text_params.font_params.font_name = (gchar *)"Serif";
text_params.font_params.font_size = 11;
text_params.font_params.font_color =
(NvOSD_ColorParams){1, 1, 1, 1};
/* Preserve original positional bounding box coordinates of detector
* in the frame so that those can be accessed after tracker */
obj_meta->detector_bbox_info.org_bbox_coords.left =
rect_params.left;
obj_meta->detector_bbox_info.org_bbox_coords.top = rect_params.top;
obj_meta->detector_bbox_info.org_bbox_coords.width =
rect_params.width;
obj_meta->detector_bbox_info.org_bbox_coords.height =
rect_params.height;
nvds_add_obj_meta_to_frame(frame_meta, obj_meta, NULL);
}
// nvds_add_frame_meta_to_batch(batch_meta, frame_meta);
}
// nvds_release_meta_lock(batch_meta);
return GST_PAD_PROBE_OK;
}
My problem :- This pipeline streammux -> nvvideoconv2 -> caps_filter -> darknet_infer -> nvvidconv -> caps_filter2 ->nvosd -> queue6 -> sink works fine, I can see bounding boxes being drawn by osd element.
But when I add tracker in the pipeline, streammux -> nvvideoconv2 -> caps_filter -> darknet_infer -> nvvidconv -> caps_filter2 ->nvtracker-> nvosd -> queue6 -> sink , I am not able to convert the gst-darknet detections to NvDsObjectMeta type.