Add support for YOLO-World

[YuhyunNa]

Hello, I appreciate your outstanding work on the DeepStream-YOLO repository, which has been incredibly helpful for using various YOLO models within the DeepStream SDK.

As the YOLO-World model, a real-time open-vocabulary object detector, has been recently released, I have extended your repository to include support for YOLO-World within the DeepStream SDK.

• Added utils/export_yoloWorld.py
  • Load YOLO-World model
  • Set custom classes (optional)
  • Modify outputs by adding a module, DeepStreamOutput()
  • Export to ONNX
• Added config_infer_primary_yoloWorld.txt
• Added docs/YOLO_World.md

[YuhyunNa]

@marcoslucianops

[lzylzylzy123456]

Have you completed the TensorRT acceleration for YOLO-World? If so, that's really great work.

[YuhyunNa]

I appreciate it!
With the DeepStream-YOLO by @marcoslucianops, I was able to make YOLO-World work within the DeepStream SDK.
The TensorRT engine building process also works fine for YOLO-World.

[lilitech1]

Solut60@icloud.com

[ZhengQuancheng]

Hello, I'm using your solution to deploy the YOLO-World model (from ultralytics' yolov8l-worldv2.pt), but I'm encountering the following issue. How can I resolve it?