The Mobile capture receipts Optical Character Recognition (MC-OCR) challenge deliver two tasks: Receipt Image Quality Evaluation and OCR Recognition. In the first task,we introduce a regression model to map various inputs such as the probability of the output OCR, cropped text boxes, images to their label. In the second task, we propose a stacked multi-model as a solution to tackle this problem. The robust models are incorporated by image segmentation, image classification, text detection, text recognition, and text classification. Follow this solution, we can get vital tackle various noise receipt types such as horizontal, skew, and blur receipt.
Dataset have 3 parts include training data, public test data and privated test data.The training set have 1,155 training examples with the respective annotated information. The public testing set consists of 391 examples without annotations. The private testing set consists of 390 examples without annotations. All of these will be found in this website (the official website of MC-OCR challenge)
Our code have to run with GPU device, thank you.
Firstly, you need to install libaries follow this code :
pip install -r requirements.txt
Then :
python3 -m pip install 'git+https://github.com/facebookresearch/detectron2.git'
- Download pretrained model from here
- Add pretrained file to folder
weights/ - Run test.py to get file results.csv
python3 test.py --folder_test [path to folder test]
