CNN_LSTM_CTC_Tensorflow
CNN_LSTM_CTC_Tensorflow copied to clipboard
Feed Images with Variable length
Hey guys i am trying to use the model to train on the images with variable length size.As you know this data set is a fixed length size and there is no problem with running the model but when it comes to other datasets such as IAM we get error since the are not fixed size.One of the techniques that have been mentioned is to do zero padding. Now my question is am i suppose to do the zero padding on the images itself before feeding them to the network and is there any other ways to overcome this issue of variable sizes.
Thanks
@watsonyanghx would you please shed some light on this? :)