search

zhiqiangdon / CU-Net

Code for "Quantized Densely Connected U-Nets for Efficient Landmark Localization" (ECCV 2018) and "CU-Net: Coupled U-Nets" (BMVC 2018 oral)
Apache License 2.0
227 stars 40 forks source link

the face landmark training script #1

Closed spark123456789 closed 5 years ago

spark123456789 commented 6 years ago

Dear Sir, Thank you for your amazing code, I wonder whether these is a training script for the face landmark, Are you planning to upload this script? Thank you for your reply.

zhiqiangdon commented 6 years ago

Sorry for the late reply. I just added the face script.

spark123456789 commented 6 years ago

thank you for your reply. Another a simple question, how can I gather the the data related to your face.json file?

1700117HLT commented 6 years ago

Dear Sir, I am learning your code. Is this required to be loaded? Can you share it? CU-Net-master/data/face_bbox.py def _compute_mean(self): meanstd_file = 'dataset/face.pth.tar'

zhiqiangdon commented 6 years ago

Hi @1700117HLT,

I find that normalizing the input images may not be necessary. The results do not show much difference.

zhiqiangdon commented 6 years ago

Hi @spark123456789,

The face data are from the 300W dataset.

1700117HLT commented 6 years ago

Dear Sir, I am learning your code face alignment,but I run the cu-net-face-prev-version,the error about preds = Evaluation.final_preds(output, center, scale, [64, 64]), I find the Evaluation.py is def final_preds(output, center, scale, res, rot): How should I define rot or can you share the new code face alignment with me ?

1700117HLT commented 5 years ago

Dear Sir, the error about File "cu-net-face-prev-version.py", line 124, in main visualizer, num_classes, flip_index) File "cu-net-face-prev-version.py", line 265, in validate preds = Evaluation.final_preds(output, center, scale, [64, 64]) TypeError: final_preds() takes exactly 5 arguments (4 given)

In the paper, rot is plus or minus 30 degrees,in Evaluation.final_preds, rot is 0? I try rot=numpy.zeros(12), but there are still many errors, then try to remove rot in final_preds(output, final_preds.py) Center, scale, res, rot), but still reported error, can you help me?

1700117HLT commented 5 years ago

I have solved the previous question by consulting my classmates, thank you for your early reply.

zhiqiangdon commented 5 years ago

You are welcome. @1700117HLT