An OCR-system based on torch using the technique of LSTM/GRU-RNN, CTC and referred to the works of rnnlib and clstm.
This work is now completely UNSTABLE, EXPERIMENTAL and UNDER DEVELOPMENT.
$ ./build.sh
- You could modify the settings in the
main.lua
directly and executeth main.lua
, the input format is clstm-like (.png
and.gt.txt
pair) and you should put all input file path in a text file. - or if you prefer to use a JSON-format configuration file, you could follow the example below, and run:
$ th main.lua -setting [setting file]
There would be a folder created in the experments
folder for every experiment. You could check out the log, settings and saved models there.
descriptions for each option could be found in main.lua
.
{
"project_name": "uy_rbm_noised",
"raw_input": false,
"hidden_size": 200,
"nthread": 3,
"clamp_size": 1,
"ctc_lua": false,
"recurrent_unit": "gru",
"test_every": 2000,
"omp_threads": 1,
"show_every": 10,
"testing_list_file": "wwr.txt",
"input_size": 48,
"testing_ratio": 1,
"max_param_norm": false,
"training_list_file": "full-train.txt",
"feature_size": 240,
"momentum": 0.9,
"dropout_rate": 0.5,
"max_iter": 10000000000,
"save_every": 10000,
"learning_rate": 0.0001,
"stride": 5,
"gpu": false,
"rbm_network_file": "rbm/wwr.rbm",
"windows_size": 10
}
BSD 3-Clause License
- Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling
- Connectionist Temporal Classification: Labelling Unsegmented Sequence Data with Recurrent Neural Networks
- RNNLIB: Connectionist Temporal Classification and Transcription Layer
- rnnlib
- clstm