hypjudy/human-pose-estimation-by-deep-learning

A simple regression based implementation/VGG16 of pose estimation with tensorflow.

human-pose-estimation-by-deep-learning's Issues

How can we generate 2d predictions on our own images?

What dataset used for training and testing?

I saw you write:

Please contact with JakeRenn for data (maybe not available currently).

And then I go to JakeRenn's project, there is no update.

I wonder what dataset is used in your project for training? Is it MPII as Wei's paper referred?

Realtime pose estimation

How do I estimate pose realtime?

The original dataset has the formatof .mat,but I have on idea of how to transform it to this porject

The original dataset has the format of .mat, but I have on idea of how to transform it to suit this porject. So can you share your training-set ?Thanks a lot !

Need help

can somebody share the complete data plz? [email protected]
thank you all!

What is the format of the annos files?

What are the joints corresponding to labels and the order of their x,y in the file?
Moreover, do the code can cope with partial labels? I.e not all joints are known or visible in every training image?
Thank you

Error when training with more than 60000 images

Hi!

I was trying to train the network with 79k images but it is showing this error

Length Error: 31

when I trained with 35k images the network was working fine. In this I have positive values for all the coordinates but in the 2nd set of images I have some negative values too . Is it because of this?

2041790.jpg -61,-461,-5,-262,-71,-252,59,-272,-18,-159,8,-333,19,-71,34,-189,32,-172,19,-113,87,-125,64,-52,67,-55,95,-83,103,-30,1
2041791.jpg -59,-461,0,-258,-68,-249,65,-267,-15,-151,9,-333,18,-71,36,-183,36,-168,25,-110,89,-121,68,-51,91,-48,96,-83,102,-13,1
2041792.jpg -60,-456,5,-252,-66,-246,75,-257,-13,-146,7,-322,27,-60,33,-189,38,-166,24,-111,89,-119,67,-51,92,-49,94,-79,103,-18,1
2041793.jpg -58,-457,5,-253,-66,-249,75,-257,-15,-150,6,-325,36,-55,32,-186,38,-166,24,-112,90,-118,67,-50,96,-51,97,-69,103,-19,1
2041794.jpg -58,-455,5,-255,-67,-253,75,-258,-15,-152,13,-297,35,-57,39,-158,36,-169,20,-115,89,-121,51,-47,98,-53,100,-28,106,-19,1
2041795.jpg -57,-452,7,-250,-62,-247,74,-254,-13,-148,3,-312,37,-57,5,-177,37,-167,21,-113,90,-119,54,-46,36,-54,79,-70,100,-35,1
2041796.jpg -52,-441,-33,-299,-93,-292,25,-306,-35,-185,69,-219,13,-98,69,-97,18,-195,7,-130,88,-140,40,-58,110,-56,75,-79,97,-31,1
2041797.jpg -56,-447,-30,-294,-85,-285,25,-302,-35,-185,70,-214,15,-98,58,-107,17,-193,2,-128,88,-142,35,-64,103,-62,82,-58,98,-31,1
2041798.jpg -57,-452,-1,-272,-74,-284,64,-261,-28,-178,50,-207,18,-75,34,-116,26,-170,14,-112,69,-108,52,-52,69,-53,108,-42,105,-28,1

I tried to change the tmp parameter in cpm.py file line 262

def sub(batch, idx):
           name = x.op.name
           tmp = x[batch, :, :, idx] * 255
           tmp = tf.expand_dims(tmp, axis = 2)
           tmp = tf.expand_dims(tmp, axis = 0)
           tf.summary.image(name + '-' + str(idx), tmp, max_outputs = 100)

even then it is showing the same error.

Could be please suggest anything regarding this?

Thanks in Advance!

Error when training with RGB images

Hi!
I'm getting this error when trying to train the model with rgb images.

Traceback (most recent call last):
  File "train.py", line 144, in <module>
    main()
  File "train.py", line 118, in main
    imgs, fm, coords, begins, filename_list = reader.get_random_batch(distort=config.toDistort)
read_data.py", line 310, in get_random_batch
    return (imgs.reshape([self.batch_size, self.img_height, self.img_width, 1]),
ValueError: cannot reshape array of size 60120000 into shape (20,1002,1000,1)

can anyone help me out with this issue?

Thanks in advance!!

Real-Time Estimation

It is really amazing project! Thanks for sharing!

I want to ask, is it good for real time pose estimations?