luoyetx/deep-landmark

Issues

• why is my caffemodel filenme “1_F_solver_iter_50000.caffemodel” but not “_iter_50000.caffemodel”?

  #36 opened 5 years ago by lmhlll
  1

• How to add one more point

  #29 opened 8 years ago by war3xxx
  2

• training 1_F network

  #3 opened 9 years ago by ikvision
  12

• When I train the model, which is entered. / Bootstrap.sh always show sh: 1: caffe not found, what is the problem? How to deal with it?

  #35 opened 7 years ago by 406baoqichao406
  5

• about test time

  #1 opened 9 years ago by yyytq
  25

• The test results of level2 and level3 are worse

  #33 opened 7 years ago by ustcluq
  2

• your paper and implementation is not the same?

  #34 opened 7 years ago by KangolHsu
  1

• [libprotobuf ERROR google/protobuf/text_format.cc:245]

  #32 opened 8 years ago by ustcluq
  2

• ImportError: No module named common

  #31 opened 8 years ago by ustcluq
  4

• not changed NM1 landmark accordingly, it is why 1_F is better?

  #28 opened 8 years ago by tengshaofeng
  2

• 2 questions about data_preprocess

  #30 opened 8 years ago by zhanpx
  2

• F1 network is not same as orignal paper

  #10 opened 8 years ago
  6

• why use the flatten layer and concat layer in level 1?

  #23 opened 8 years ago by Dan1900
  2

• where is the functools？

  #24 opened 8 years ago by wanglu0316
  0

• About how you preprocess the image

  #27 opened 8 years ago by breadbread1984
  1

• sh: 1: caffe: not found

  #26 opened 8 years ago by NIEYALI
  1

• Why some landmarks failed?

  #25 opened 8 years ago by hzh8311
  1

• Hard to repeat level3's mean error at some points

  #14 opened 8 years ago by mzpmzk
  6

• loss EuclideanLoss can change into SoftmaxWithLoss？thanks

  #22 opened 8 years ago by mttsky
  1

• The test with the landmark.py you applied is not accurate.

  #16 opened 8 years ago by XiaXuehai
  6

• hello why twice use np.random.rand() > 0.5 in level1 generate_hdf5 .thanks

  #21 opened 8 years ago by mttsky
  4

• pre-trained models?

  #20 opened 8 years ago by iqbalnaved
  1

• compilation issue because of looped dependencies

  #18 opened 8 years ago by saikrishna-1996
  1

• flip() function

  #19 opened 8 years ago by saikrishna-1996
  1

• Why there is a final relu in the 1_F network

  #17 opened 8 years ago by laoreja
  1

• How to set the host value when testing?

  #15 opened 8 years ago by XiaXuehai
  2

• A question about face detection in README

  #13 opened 8 years ago by xvbw
  2

• a question about hdf5

  #12 opened 8 years ago by yyytq
  0

• Hard to repeat the training loss or accuracy as your trained model

  #11 opened 9 years ago by wenston2006
  3

• how to solve the face just on the edge of the image?

  #8 opened 9 years ago by wuqiangch
  2

• Training and testing procedures are not the same

  #9 opened 9 years ago by wuqiangch
  3

• why the results are not stable？

  #7 opened 9 years ago by wuqiangch
  3

• how to check the hd5 data ?

  #6 opened 9 years ago by wuqiangch
  18

• Model blending using only 1_F is better than using all three networks in level1.

  #5 opened 9 years ago by mariolew
  1

• About the modified caffe

  #2 opened 9 years ago by mariolew
  1