SynthText

This a modified version of Ankush's code for generating synthetic text images which support right-to-left languages such as Persian and Arabic.

As in Persian and Arabic, words are made using connected letters, this code only outputs the word level bounding boxes.

A sample output of the code

Install the necessary libs and create a conda environment

First install the fribidi library

sudo apt install libfribidi-dev

Then create a conda environment from the environment.yml file:

conda env create -f environment.yml

Activate the environment with the following command:

conda activate gentext

The rest of the README is from the original repository

Code for generating synthetic text images as described in "Synthetic Data for Text Localisation in Natural Images", Ankush Gupta, Andrea Vedaldi, Andrew Zisserman, CVPR 2016.

Synthetic Scene-Text Image Samples

The library is written in Python. The main dependencies are:

pygame, opencv (cv2), PIL (Image), numpy, matplotlib, h5py, scipy

Generating samples

python gen.py --viz

This will download a data file (~56M) to the data directory. This data file includes:

dset.h5: This is a sample h5 file which contains a set of 5 images along with their depth and segmentation information. Note, this is just given as an example; you are encouraged to add more images (along with their depth and segmentation information) to this database for your own use.
data/fonts: three sample fonts (add more fonts to this folder and then update fonts/fontlist.txt with their paths).
data/newsgroup: Text-source (from the News Group dataset). This can be subsituted with any text file. Look inside text_utils.py to see how the text inside this file is used by the renderer.
data/models/colors_new.cp: Color-model (foreground/background text color model), learnt from the IIIT-5K word dataset.
data/models: Other cPickle files (char_freq.cp: frequency of each character in the text dataset; font_px2pt.cp: conversion from pt to px for various fonts: If you add a new font, make sure that the corresponding model is present in this file, if not you can add it by adapting invert_font_size.py).

This script will generate random scene-text image samples and store them in an h5 file in results/SynthText.h5. If the --viz option is specified, the generated output will be visualized as the script is being run; omit the --viz option to turn-off the visualizations. If you want to visualize the results stored in results/SynthText.h5 later, run:

python visualize_results.py

Pre-generated Dataset

A dataset with approximately 800000 synthetic scene-text images generated with this code can be found here.

Adding New Images

Segmentation and depth-maps are required to use new images as background. Sample scripts for obtaining these are available here.

predict_depth.m MATLAB script to regress a depth mask for a given RGB image; uses the network of Liu etal. However, more recent works (e.g., this) might give better results.
run_ucm.m and floodFill.py for getting segmentation masks using gPb-UCM.

For an explanation of the fields in dset.h5 (e.g.: seg,area,label), please check this comment.

Pre-processed Background Images

The 8,000 background images used in the paper, along with their segmentation and depth masks, have been uploaded here: http://zeus.robots.ox.ac.uk/textspot/static/db/<filename>, where, <filename> can be:

imnames.cp [180K]: names of filtered files, i.e., those files which do not contain text
bg_img.tar.gz [8.9G]: compressed image files (more than 8000, so only use the filtered ones in imnames.cp)
depth.h5 [15G]: depth maps
seg.h5 [6.9G]: segmentation maps

Note: I do not own the copyright to these images.

Generating Samples with Text in non-Latin (English) Scripts

@JarveeLee has modified the pipeline for generating samples with Chinese text here.

Further Information

Please refer to the paper for more information, or contact me (email address in the paper).

adavoudi / SynthText