Jiancheng Pan
Jiancheng Pan
> > > @Qia98 , when evaluating , are the predicted labels correct compared to the gt ? or logical ? I printed out the labels , it never matches...
将shard_dis_MSSF改成shard_dis_SWAN即可，这是是一个方法，最终的时候删除错误了。
感谢您的细心指出，关于RSITMD在论文中时出现的错误：实际上应该是856 validation images, and 452 test images。在后续工作中将会订正。
We add the inference examples. If you have any questions, we welcome you to propose issues.
1. LAE-DINO uses pretrained GroundingDINO for faster convergence—you may also try training from scratch. 2. Convergence is judged by the validation set, with a large training epoch setting. Using 32...
Hi, thanks. Training on diverse categories may reduce recognition performance for certain classes. LAE-1M is designed to enhance recognition across diverse categories within a practical scope. For improved performance on...
We have released the json file in LAE-1M dataset On Mon, Apr 21, 2025 at 10:06 hyeonsu-hwang ***@***.***> wrote: > Thank you! > I'd really appreciate it if you could...
you can find it from baidu disk link. On Tue, Apr 29, 2025 at 10:52 one23sunnyQQ ***@***.***> wrote: > *one23sunnyQQ* left a comment (jaychempan/LAE-DINO#15) > > > 您好，您能通过邮件提供一下processed_DOTAv2_train.json 和processed_DIOR_train.json >...
> Hi, I am also interested in `processed_DOTAv2_train.json` and `processed_DIOR_train.json` files and was wondering if they will be released in the OneDrive as well? We will update the OneDrive, or...
``torch>=2.3.1 and deepspeed>=0.17.4`` is also ok.