search

arundasan91 / DeXpression

TFLearn Implementation of DeXpression architecture. Batch normalization is used instead of LRN. Gives a precision of 99.3 percent, recall of 99.2 percent and f1-score of 99.2 percent on CKPlus Dataset for human emotion recognition from frontal facial images.
13 stars 8 forks source link

numpy to feed in network #3

Closed NomiMalik0207 closed 6 years ago

NomiMalik0207 commented 6 years ago

Cannot feed value of shape (350,) for Tensor 'TargetsData/Y:0', which has shape '(?, 7)' when i am trying to feed numpy into network is give the above error? how i can solve this. thanks

arundasan91 commented 6 years ago

Your output label file is not of proper shape. Required shape is (num_images, 7) where num_images are the number of images in training set.

Please check the shape of your output y values by using for example variable_y.shape

NomiMalik0207 commented 6 years ago

yup.. its shape is (1304,1) i can reshape it using onehot encoding to (1304,8) but when i try to change (1304,7) it give the error of index 7 is out of bounds for axis 0 with size 7. this have you anyother script of changing dataset in to numpy? if yes than please share the link it will be a greatness of your's . thank you

arundasan91 commented 6 years ago

I see that your numpy file has 8 labels. Please change the final softmax layer of the network to reflect that. You could just change https://github.com/arundasan91/DeXpression/blob/master/model_deXpression.py#L26 and try again.