search

YangZeyu95 / unofficial-implement-of-openpose

Implement of Openpose use Tensorflow
272 stars 79 forks source link

my own dataset #40

Open soveidadelgarmi opened 5 years ago

soveidadelgarmi commented 5 years ago

Hi, when I train openpose on my own dataset(small dataset), how to know it works well for my dataset in data augmentation part of get_data() in pose_dataset.py (lines 355-359 )? is it possible that new coordinates of key points don't adjust well on the augmented image? or it works correctly for any dataset? because after running run.py with my pre_trained model, the output image of heatmap and vector map are not acceptable.is it ok because of my small dataset or I should check the augmentation part? Can someone help me, Thank you.

ds = MapDataComponent(ds, pose_random_scale) ds = MapDataComponent(ds, pose_rotation) ds = MapDataComponent(ds, pose_flip) ds = MapDataComponent(ds, pose_resize_shortestedge_random) ds = MapDataComponent(ds, pose_crop_random)

YangZeyu95 commented 5 years ago

You can check the ground truth in tensorboard. Data argument part works well in all my dataset.

soveidadelgarmi commented 5 years ago

Thank you for your response. I use this code in colab for tensorborad. is it correct? !tensorboard --logdir './checkpoints/train/2019-7-21-5-2-28/model-210' TensorBoard 1.14.0 at http://15e3afa71241:6006/ (Press CTRL+C to quit)

then I click on this URL but this page opened...The requested URL could not be retrieved thanks.

YangZeyu95 commented 5 years ago

just run ’tensorboard --logdir checkpoints/train/2019-7-21-5-2-28‘ at the root path of the project

soveidadelgarmi commented 5 years ago

@YangZeyu95 I couldn't see the ground truth in tensorboard. just the graph is loaded also I can't see the images in tensorboard on google colab. Is there another way to check ground truth? I have another question, in pose_dataset.py, in line 86, what is transform? I searched but I don't find anything. transform = list(zip( [1, 6, 7, 9, 11, 6, 8, 10, 13, 15, 17, 12, 14, 16, 3, 2, 5, 4], [1, 7, 7, 9, 11, 6, 8, 10, 13, 15, 17, 12, 14, 16, 3, 2, 5, 4] ))

soveidadelgarmi commented 5 years ago

Hi, I would like to ask what is the function of 'peaks' in the script of 'process_paf'. I see the code like this: pafprocess.process_paf(peaks, heat_mat, paf_mat) There is already a 'heat_mat'. Why still need 'peaks'? Thanks.

Clarence-wen commented 5 years ago

@YangZeyu95 I couldn't see the ground truth in tensorboard. just the graph is loaded also I can't see the images in tensorboard on google colab. Is there another way to check ground truth? I have another question, in pose_dataset.py, in line 86, what is transform? I searched but I don't find anything. transform = list(zip( [1, 6, 7, 9, 11, 6, 8, 10, 13, 15, 17, 12, 14, 16, 3, 2, 5, 4], [1, 7, 7, 9, 11, 6, 8, 10, 13, 15, 17, 12, 14, 16, 3, 2, 5, 4] ))

Did you konw the transform's mean?