idea-research / grounding-dino-1.5-api Goto Github PK

看了一下paper，GD1.5版本的tensorrt推理只有fp32的，是因为有的算子还不支持fp16的吧？

May I somehow be provided the model file to deploy it locally and maybe fine-tune it for a more specific task?

So far I only play with the API token, I guess would be from Pro version, may I know if I can call Edge one?

我想批量跑500张图，测试一下效果。发现读取文件夹内500张图，只能跑5张图。不是申请到的Token 有200 次 免费 API 调用么 谢谢

Why does not perform the gradio_app.py program by GPUs. It doesn't produce the predicted result. How to resolve it. Thanks a lot.

Hi!
Congrats on Grounding DIO 1.5, really cool work 🔥🚀
It would be great to claim your paper on our Daily Paper Page. There are already comments on the paper threads : )
Feel free to engage with the community and to get more feedbacks!

Thank you for your great work!

Is there any plan to release the model checkpoint/weight files?

Hi, thanks for sharing the excellent work !!!

When runing the demo using python demo.py --token https://deepdataspace.com/request_api, here raises an error as follows:

File "/Grounding-DINO-1.5-API-master/gdino/model_wrapper.py", line 124, in get_image_url
url = self.client.upload_file(image)
File "/anaconda3/envs/GDINO/lib/python3.9/site-packages/dds_cloudapi_sdk/client.py", line 68, in upload_file
assert rsp.status_code == 200
AssertionError

Could you tell me how to fix the promblem?
Thanks in advance.

Hi! When I run gradio_app.py, I found an error as illustrated in the above figure.

Thanks for your great work!
I'd like to ask how to use the video object detection with GroudingDINO v1.5 pro, or if you are going to release the demo for it?

https://deepdataspace.com/request_api this link at least doesn't work in US now.