search

AlexanderParkin / ChaLearn_liveness_challenge

ChaLearn Face Anti-spoofing Attack Detection Challenge@CVPR2019
MIT License
407 stars 91 forks source link

Can Anyone run success this project? #9

Open John1231983 opened 4 years ago

John1231983 commented 4 years ago

I think the author has not double check the repo. Many error and missing file in this repo. If someone has sucess run it, please let me know

xudou3 commented 4 years ago

yeah, it will need to comment some parts for missing file and i changed some setting in opt file so it can run on my single 1070

mrkstt commented 4 years ago

Hi @John1231983 I also successfully run the code. Probably this note can be useful:

  1. Remark all mobilenetv2 import (there are three files, find using "find in file" option on visual code)

  2. Make sure all paths are correct, including "data_root" at init_dataloader.py, weights paths at resnetcaffe*.py, and etc.

  3. Remark this part on "trainer.py": """ self.visdom_log_file = os.path.join(self.opt.out_path, 'log_files', 'visdom.log') self.vis = Visdom(port = opt.visdom_port, log_to_filename=self.visdom_log_file, env=opt.expname + '' + str(opt.fold))

    self.vis_loss_opts = {'xlabel': 'epoch', 
                      'ylabel': 'loss', 
                      'title':'losses', 
                      'legend': ['train_loss', 'val_loss']}

self.vis_tpr_opts = {'xlabel': 'epoch', 
                      'ylabel': 'tpr', 
                      'title':'val_tpr', 
                      'legend': ['tpr@fpr10-2', 'tpr@fpr10-3', 'tpr@fpr10-4']}

self.vis_epochloss_opts = {'xlabel': 'epoch', 
                      'ylabel': 'loss', 
                      'title':'epoch_losses', 
                      'legend': ['train_loss', 'val_loss']} """

  4. Don't forget to download the model*.pth from CASIA-SURF dataset, and put on the corresponding folder as shown on point 5 (pth path).

  5. I run the val script as follows:

python inference.py --config /home/face-spoofing/ChaLearn_liveness_challenge/data/opts/exp1_2stage_seed1/exp1_2stage_seed1_fold1_stage1.opt --pth /home/face-spoofing/ChaLearn_liveness_challenge/data/opts/exp1_2stage_seed1/fold1/checkpoints/model_30.pth --input_list /home/face-spoofing/ChaLearn_liveness_challenge/data/lists/folds_by_fakes/fold1/val_list.txt --output_list /home/face-spoofing/ChaLearn_liveness_challenge/output.txt

Cheers!

harisudhan312 commented 4 years ago

@AlexanderParkin

Hi,

I found the links for exp1_2stage and exp3c on google drive to be broken. Can the author check and fix the broken links?

zhangxiaopang88 commented 4 years ago

Hello, download the model in the fourth article you wrote. Where did you download the model? could you tell me the specific path? @mrkstt

cakuba commented 4 years ago

Hi, could you please kindly tell me where I can find "the model*.pth from CASIA-SURF dataset" as you mentioned? @mrkstt I cannot locate it anywhere within CASIA-SURF dataset, or I am missing something here...

mrkstt commented 4 years ago

I think the authors removed the .pth model file. If you're interested, please drop an email to me kahlil@ieee.org @cakuba @zhangxiaopang88

Thanks and sorry for the late reply.