not working for below type of images (other than given by you). I think we need to put images in particular format

[harvardnlp]

not working for below type of images (other than given by you). I think we need to put images in particular format

[vyaslkv]

for this also not working

giving results as
\begin{array} { r c l } \end{array}

[vyaslkv]

I have even done the preprocessing step but still not getting results:
onmt_preprocess -data_type img
-src_dir data/im2text/images/
-train_src data/im2text/src-train.txt
-train_tgt data/im2text/tgt-train.txt -valid_src data/im2text/src-val.txt
-valid_tgt data/im2text/tgt-val.txt -save_data data/im2text/demo
-tgt_seq_length 150
-tgt_words_min_frequency 2
-shard_size 500
-image_channel_size 1

[vyaslkv]

is there any way I could convert my images to your type of images

[vyaslkv]

If I need to retrain the model for my type of images then can you tell me the format it will be required

format of data to retrain the model using python

[vyaslkv]

is there any dataset which I could use with the python implementation to train the model and get the results on normal other images as well

[vyaslkv]

http://bit.ly/2E7H8tX

I found the above data but it is in different format can you help me how can I use this type of data for training

[da03]

Sorry for the delay. Are you training a new model or are you using the pretrained one? The pretrained one is only trained on a particular rendering setup so it's not surprising that it fails on out-of-training-distribution data at test time. To make it work, a new model needs to be trained with data encompassing enough test-style images (font, size, etc).

[vyaslkv]

I used the pretrained model. Thanks for replying!!