search

mit-han-lab / once-for-all

[ICLR 2020] Once for All: Train One Network and Specialize it for Efficient Deployment
https://ofa.mit.edu/
MIT License
1.87k stars 334 forks source link

Applying this technique to detection problem (training on the COCO detection dataset)? #26

Open edwardnguyen1705 opened 4 years ago

edwardnguyen1705 commented 4 years ago

Dear @han-cai ,

Thanks for sharing your excellent work. I am just curious about how we can apply this technique to detection problem (training on the COCO detection dataset)? Appreciate your time.

darrenzhang1007 commented 3 years ago

I also want to know eagerly. Thanks

hangxu124 commented 3 years ago

hi @edwardnguyen1705 @DarrenmondZhang ,I'd like to know if you have some trials in detection tasks.. I‘m now trying to do something use ofa in detection..

edwardnguyen1705 commented 3 years ago

@hangxu124 I have not tried this so far. Hope having time to try it later.

hangxu124 commented 3 years ago

@hangxu124 I have not tried this so far. Hope having time to try it later.

ok, I'm now working on it. Training it from scratch on imagenet, finetune supernet in detection then finetune the subnet

slimwangyue commented 3 years ago

@hangxu124 I have not tried this so far. Hope having time to try it later.

ok, I'm now working on it. Training it from scratch on imagenet, finetune supernet in detection then finetune the subnet

May I ask how is your training going?

Darshcg commented 3 years ago

Hi @edwardnguyen1705 @hangxu124,

Did you try Implementing OFA for Detection? I also wanted to do the same, needed Guidance.

Thanks

timatom commented 3 years ago

I would also like to see how Han's lab produced object detection, as this will help further research in embedded ML.

I've recently seen Han's lab share person detection and face/mask detection demonstrations a few months ago.

@han-cai, if there are any recommendations you have in reproducing these results, it would be greatly appreciated. Your lab's work is very impressive.

Jon-drugstore commented 2 years ago

Dear @han-cai ,

Thanks for sharing your excellent work. I am just curious about how we can apply this technique to detection problem (training on the COCO detection dataset)? Appreciate your time.

Did you try Implementing OFA for Detection? Thanks

Xingxu1996 commented 2 years ago

mark，I want to apply it to detection, too. Fix detection head and find a backbone through once-for-all?