YOLO-World
YOLO-World copied to clipboard
Can you give a simple example to test on an image?
Can you give a simple example to test on an image?
Hi @sssss99999, sure, we will do it as soon as possible. You can try our demo first.
A simple example without gradio integration would really help.
Hi @aqibsaeed, we have updated YOLO-World with an image-demo
, you can have a try~
Neat! Thanks!
I notice it requires passing an image path, why not have a support for image array (e.g., directly from cv2) here: https://github.com/AILab-CVC/YOLO-World/blob/e4e5fc4119416c12c79e76580f226a442d4dca72/image_demo.py#L69-L71
Can we directly pass image data here: https://github.com/AILab-CVC/YOLO-World/blob/e4e5fc4119416c12c79e76580f226a442d4dca72/image_demo.py#L71?
It would be useful for live detection.
Is it possible to send an example for test_config? THANKS
I notice it requires passing an image path, why not have a support for image array (e.g., directly from cv2) here:
https://github.com/AILab-CVC/YOLO-World/blob/e4e5fc4119416c12c79e76580f226a442d4dca72/image_demo.py#L69-L71
Can we directly pass image data here:
https://github.com/AILab-CVC/YOLO-World/blob/e4e5fc4119416c12c79e76580f226a442d4dca72/image_demo.py#L71
? It would be useful for live detection.
you can save the image array as a tmp file, and pass the tmp path into the dict. refer to https://github.com/AILab-CVC/YOLO-World/blob/24c7121cf83c5808efef91c91b76277980feb99e/demo.py#L62
I notice it requires passing an image path, why not have a support for image array (e.g., directly from cv2) here: https://github.com/AILab-CVC/YOLO-World/blob/e4e5fc4119416c12c79e76580f226a442d4dca72/image_demo.py#L69-L71
Can we directly pass image data here: https://github.com/AILab-CVC/YOLO-World/blob/e4e5fc4119416c12c79e76580f226a442d4dca72/image_demo.py#L71
? It would be useful for live detection.
you can save the image array as a tmp file, and pass the tmp path into the dict. refer to
https://github.com/AILab-CVC/YOLO-World/blob/24c7121cf83c5808efef91c91b76277980feb99e/demo.py#L62
same question it is too slow
Hi, all (@trihook, @mingxingpu, @taltlusty, @aqibsaeed, @sssss99999 ), this problem has been resolved and I've provided a pretty simple demo code to do this. Using an image numpy array is fine. You can find it at demo/simple_demo.py
.