a question about 'ifp import ifp_sample' about se-ssd

[Vegeta2020]

Thanks for you feedback. It's an issue about ifp, you may use git clone https://github.com/jackd/ifp-sample.git; pip install -e ifp-sample to install this module. I also updated README to avoid such a problem.

[qinhao1994]

Thank you for your reply！I have solved this problem.
But I ran into a new problem when I tried to run train:
model_ema = copy.deepcopy(model)
TypeError: can't pickle _thread.RLock objects
i have set workers_per_gpu to 0
Looking forward to your reply again

[lijin2021]

I also encountered this problem. I created a model_ema repeatedly in train.py to replace its original deepcopy method. After the replacement, the model can be trained normally.

[Vegeta2020]

@qinhao1994 I'm not sure the reason for this bug, as it could be run on my machine. And I googled it, the results show it might be solved by correcting parameter setting, perhaps you may try it first. @lijin2021 I tried to avoid repeat copy to keep higher efficiency.