search

liuwei16 / CSP

High-level Semantic Feature Detection: A New Perspective for Pedestrian Detection, CVPR, 2019
754 stars 191 forks source link

Can not reproduce the results in the paper！ #45

Open a1aaaaagithub opened 5 years ago

a1aaaaagithub commented 5 years ago

Can not reproduce the results in the paper, using resnet-50, I loaded the weight provided by the author (net_e110_l0.0082234138241.hdf5), in the parse_det_offset function, when the threshold corresponding to seman is 0.01, there are countless misdetections, small targets Pedestrians can hardly detect it. Can anyone tell me how to reproduce the results in the author's paper?

msha096 commented 5 years ago

check this https://github.com/liuwei16/CSP/issues/38

super-wcg commented 5 years ago

@a1aaaaagithub did you solve the problem? I run his code and his model with paper's setting,but I can't get 4.5(MR), Do you know the reason?

super-wcg commented 5 years ago

@a1aaaaagithub did you solve the problem? I run his code and his model with paper's setting,but I can't get 4.5(MR), Do you know the reason?

I run test model on Caltech and use his data/cache file.

fyture commented 3 years ago

hello？Did you evaluate this weight？I evaluated this weight and got the MR of 99.8%. I am confused and do not know what's wrong.

fyture commented 3 years ago

@a1aaaaagithub did you solve the problem? I run his code and his model with paper's setting,but I can't get 4.5(MR), Do you know the reason?

I run test model on Caltech and use his data/cache file.

What result do you got? I got MR of 99.8% which seems not right.