Skip to content

Latest commit

 

History

History
48 lines (44 loc) · 2.25 KB

finetune.md

File metadata and controls

48 lines (44 loc) · 2.25 KB
Raw

Fine-tuning MAERec

We build MAERec with MMOCR-1.0.

1. Installation

cd mmocr-dev-1.x
conda create -n mmocr1.0 python=3.8 -y
# PyTorch 1.6 or higher is required
pip install torch==1.8.1+cu111 torchvision==0.9.1+cu111 -f https://download.pytorch.org/whl/torch_stable.html
pip install -U openmim
mim install mmengine
mim install mmcv
mim install mmdet
pip install timm
pip install -r requirements/albu.txt
pip install -r requirements.txt
pip install -v -e .

2. Fine-tuning MAERec on Union14M-L

    1. Download the pre-trained ViT
    1. Modify the config file mmocr-dev-1.x/configs/textrecog/maerec/maerec_b_union14m to set the pretrained field in model.backbone to the path of the pre-trained ViT model.
    1. Modify the config file mmocr-dev-1.x/configs/textrecog/_base_/datasets/Union14M_train.py to set the union14m_data_root field to the path of the Union14M-L dataset.
    1. Modify the config file mmocr-dev-1.x/configs/textrecog/_base_/datasets/Union14M_benchmark.py to set the union14m_root field and union14m_benchmark_root to the path of the Union14M-Benchmarks.
    1. Run the following command to fine-tune MAERec on Union14M-L.
    cd mmocr-dev-1.x
# training with single GPU
python tools/train.py configs/textrecog/maerec/maerec_b_union14m.py
# training with multiple GPUs (8 GPUs in this example)
bash tools/dist_train.sh configs/textrecog/maerec/maerec_b_union14m.py 8

3. Evaluate MAERec on Union14M-Benchmarks

    1. Download the pre-trained MAERec
    1. Modify the config file mmocr-dev-1.x/configs/textrecog/_base_/datasets/Union14M_benchmark.py to set the union14m_root field and union14m_benchmark_root to the path of the Union14M-Benchmarks.
    1. Run the following command to evaluate MAERec on Union14M-Benchmarks.
    cd mmocr-dev-1.x
# evaluation with single GPU
python tools/test.py configs/textrecog/maerec/maerec_b_union14m.py \
    {PATH TO PRETRAINED MAEREC} \
# evaluation with multiple GPUs (8 GPUs in this example)
bash tools/dist_test.sh configs/textrecog/maerec/maerec_b_union14m.py \
    {PATH TO PRETRAINED MAEREC} \
    8 \