instructions.txt

#Python-3.7.5

pip install -r requirements.txt

To generate annotations for image dir (saves "<image_filename> <predicted_text>" into the output file):
./my_run.sh <Image_Dir> <Output_File_Path>

To run the web app
1. Using default model:
python app.py ./exported-model/default-model
2. Using modified model:
python app.py ./exported-model/custom-model


=================================================

TO TEST ALL FILES AND PROCEDURES
--------------------------------

pip install -r requirements.txt
pip install -r ./text_renderer/requirements.txt
pip install ./attention-ocr/


To test model-checkpoints:
--------------------------

1. To test default model:
	1.1. Add the checkpoint file path to the "checkpoint" file in model-path
eg. model_checkpoint_path: "model.ckpt-8200"
must be present in checkpoint file in <checkpoints_dir>
	1.2. aocr test --target-embedding-size 20 --use-gru  --model-dir <default_model_checkpoints_dir> <test_filename>.tfrecords

2. To test custom model:

	2.1. Add the checkpoint file path to the "checkpoint" file in model-path
eg. model_checkpoint_path: "model.ckpt-24960"

	2.2. aocr test --target-embedding-size 20 --custom-cnn --model-dir <xustom_model_checkpoints_dir> <test_filename>

eg. aocr test --target-embedding-size 20 --custom-cnn --model-dir ./checkpoints ./datasets/new-test.tfrecords 


To generate images
------------------

1. Make changes to text_renderer/generate_images.sh according to number of images to generate.
2. Run text_renderer/generate_images.sh
3. The images will be saved in text_renderer/output/default

To generate tfrecords
---------------------

aocr dataset <annotations_file_path> <dataset_filename_to_save.tfrecords>