Closed syssj1002 closed 2 years ago
I want to know too!!
i found plot_dual_img
function in the code, but struggling with usage of it (no example in code at all).
So I am wondering too
I also didn't find a visualization of outputting a single image, but I referenced DETR's visualization code, which can then be used directly here.
can you send the your code to me?thanks a lot!
I also didn't find a visualization of outputting a single image, but I referenced DETR's visualization code, which can then be used directly here. Are you using DAB-Deformable-DETR, can you send me the reasoning visualization file?
I also want to know! Somebody helps!
Hello everyone, I provide a note for a demo of inference and visualization in https://github.com/IDEA-opensource/DAB-DETR/blob/main/inference_and_visualize.ipynb. Hope the file is helpful to you. Thanks for your suggestions.
Hello everyone, I provide a note for a demo of inference and visualization in https://github.com/IDEA-opensource/DAB-DETR/blob/main/inference_and_visualize.ipynb. Hope the file is helpful to you. Thanks for your suggestions.
Hello ,when I use your inference code, when I get to this line of code:
"output = model(image[None]) output = postprocessors['bbox'](output, torch.Tensor([[1.0, 1.0]]))[0]"
and then, how should I solve this problem, please tell me more , Thanks a lot!
Hello everyone, I provide a note for a demo of inference and visualization in https://github.com/IDEA-opensource/DAB-DETR/blob/main/inference_and_visualize.ipynb. Hope the file is helpful to you. Thanks for your suggestions.
Hello ,when I use your inference code, when I get to this line of code:
"output = model(image[None]) output = postprocessors['bbox'](output, torch.Tensor([[1.0, 1.0]]))[0]"
and then, how should I solve this problem, please tell me more , Thanks a lot!
Our code is running on CPUs, whereas the dab_deformable_detr supports GPU only. You can send images and models to GPU before inference. A simple way to modify it:
model = model.cuda()
image = image.cuda()
output = model(image[None])
Hello everyone, I provide a note for a demo of inference and visualization in https://github.com/IDEA-opensource/DAB-DETR/blob/main/inference_and_visualize.ipynb. Hope the file is helpful to you. Thanks for your suggestions.
Hello ,when I use your inference code, when I get to this line of code: "output = model(image[None]) output = postprocessors['bbox'](output, torch.Tensor([[1.0, 1.0]]))[0]" and then, how should I solve this problem, please tell me more , Thanks a lot!
Our code is running on CPUs, whereas the dab_deformable_detr supports GPU only. You can send images and models to GPU before inference. A simple way to modify it:
model = model.cuda() image = image.cuda() output = model(image[None])
Thank you very much for the answer, I also solved the problem, wish you success in your career, thanks again
I find this code will inference validation dataset,but I don‘t konw how to inference local pictures.
Use PIL Image with path to your image and None as target
I update the notebook to support visualizing custom images. See https://github.com/IDEA-opensource/DAB-DETR/blob/main/inference_and_visualize.ipynb for more details.
I find this code will inference validation dataset,but I don‘t konw how to inference local pictures.
Thanks for your great work. After testing the code, how can I print the image results with bounding box? Is there any command or code?