xialeiliu / crowdcountingcvpr18 Goto Github PK

Hello xialeiliu,

what pre-processing we have to do for this model?can u provide the mean value for mean subtraction?

regards

Hello xialeiliu,
Is it possible to get the people count from the last conv layer ? if yes how?

regards

hi,i have a few question.
1)what is the ranking picture label?is the ranking level?like 1,2,3?represents the size of the croped picture?
2)in Fig3,when the N=1,what is the 11(N1)in the bule network lasts parts mean?the number of the ranking picture?or hat_Ci?
3)in the paper,you said that

In practice we sample minibatches of 25 images which contain 5 sets of 5 images which can be compared among them resulting in a total of 5 × (4 + 3 + 2 + 1) = 50 pairs in one minibatch.

what is the (4+3+2+1) mean?

thank you so much!!

Hi, thanks for your awesome work! I am very interested in the two unlabeled datasets collected by you in the paper.

Can you share the Unlabeled Dataset? Not the gt dataset. The dataset you collected from google.

Thanks for your nice job!@xialeiliu
I did not find the training code, could you please tell me how to train the model and reproduce the results of your paper?
Thanks for your reply!