mahendra047/craft-remade

Implementation of CRAFT Text Detection

Re-Implementing CRAFT-Character Region Awareness for Text Detection

Objective

• Reproduce weak-supervision training as mentioned in the paper https://arxiv.org/pdf/1904.01941.pdf
• Generate character bbox on all the popular data sets.
• Expose pre-trained models with command line interface to synthesize results on custom images

Cone the repository

git clone https://github.com/autonise/CRAFT-Remade.git
cd CRAFT-Remade

Option 1: Conda Environment Installation

conda env create -f environment.yml
conda activate craft

Option 2: Pip Installation

pip install -r requirements.txt

Running on custom images

Put the images inside a folder.
Get a pre-trained model from the pre-trained model list (Currently only strong supervision using SYNTH-Text available)
Run the command -

python main.py train_synth --mode=synthesize --model=./model/final_model.pkl --folder=./input

Pre-trained models

Strong Supervision

SynthText - https://drive.google.com/open?id=1qnLM_iMnR1P_6OLoUoFtrReHe4bpFW3T

Weak Supervision

• ICDAR 2013 - In Progress
• ICDAR 2015 - yet_to_be_completed
• ICDAR 2017 - yet_to_be_completed
• Total Text - yet_to_be_completed
• MS-COCO - yet_to_be_completed

Pre-generated on popular data sets

• ICDAR 2013 - In Progress
• ICDAR 2015 - yet_to_be_completed
• ICDAR 2017 - yet_to_be_completed
• Total Text - yet_to_be_completed
• MS-COCO - yet_to_be_completed

How to train on your own dataset

Download the pre-trained model on Synthetic dataset at https://drive.google.com/open?id=1qnLM_iMnR1P_6OLoUoFtrReHe4bpFW3T
Make your own custom dataloader as in train_weak_supervision/dataloader.DataLoaderMIX
Run the command -

python main.py weak_supervision --model=/path/to/pre-trained/Synth-Text-Model --iterations=epochs-of-weak-supervision