How to add a new element to NvDsObjectMeta?

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

• Hardware Platform (Jetson / GPU): GPU
• DeepStream Version: 5.1
• Issue Type( questions, new requirements, bugs): questions.

Hi everyone, I’m just a newbie to deepstream.

I ran the deepstream-ssd-parser with KLT tracker of deepstream-nvanalytics. According to my knowledge, the NvDsObjectMeta includes class_id, objec_id, bounding_box,… Can I add a new element to NvDsObjectMeta?. Specifically, I wanna add midpoint of object in order to display it with nvosd.

Could you guys give me some comments or advice about that? Thanks in advanced.

3

If you just want to display the midpoint, you can refer deepstream-test1 for how to add display meta

4

Thanks, I’ll notice that. Btw, how to add the midpoint to the NvDsObjectMeta? Because I need it for post-processing as well.

5

I didn’t get, you can calculate the mid point if already get the bbox

6

Yeah exactly. My question now is how to change or custom the NvDsObjectMeta in python ? Is it possible ? I also wanna draw a circle at the midpoint, how can I do it ?

7

Osd is used to do the real draw work, so you just add the display meta for circle and osd will help you do that, refer Deepstream Python API Reference — Deepstream Deepstream Version: 6.1.1 documentation

9

Hi, I followed your instruction and it didn’t work. Here is my code:

rect_para = obj_meta.rect_params
circle_meta = pyds.nvds_acquire_display_meta_from_pool(batch_meta)
circles = circle_meta.circle_params
circle = circles[0]
circle.has_bg_color = 1
circle.bg_color.set(1.0, 1.0, 0.0, 1.0)
circle.radius = 5
circle.xc = int(rect_para.left + int(rect_para.width / 2))
circle.yc = int(rect_para.top + int(rect_para.height / 2))
circle.circle_color.set(1.0, 1.0, 0.0, 1.0)
pyds.nvds_add_display_meta_to_frame(frame_meta, display_meta)

Could you give me some advise or comments about what’s wrong here ? Thank you very much.

11

Can you refer deepstream_tao_apps/apps/tlt_others/deepstream-bodypose2d-app at release/tlt3.0 · NVIDIA-AI-IOT/deepstream_tao_apps · GitHub