Skip to content

Latest commit

 

History

History
23 lines (13 loc) · 1.23 KB

README.md

File metadata and controls

23 lines (13 loc) · 1.23 KB
Raw

CBS Sub-Character Component Recognition

This directory contains the code that implements the CBS (Chu Bamboo Slip) sub-character component recognition task. Given an image of a CBS character, the objective is the recognize the components that make up the character. This step is useful for handling CBS characters that do not have a modern Chinese equivalence. This task is framed as a multi-label classification task.

Data

Before running the code, you need to download the CHUBS data from 🤗 Datasets: https://huggingface.co/datasets/chen-yingfa/CHUBS

Model

For ResNet, we fine-tune the resnet50 provided by the PyTorch library. For ViT, we fine-tune the public model vit_base_patch16_224_in21k at 🤗 HuggingFace, which was pre-trained on ImageNet.

Training

Execute the following to train either of the baseline models using default arguments.

  • For training ViT: python train_vit.py --data_dir path/to/data
  • For training ResNet: python train_resnet.py --data_dir path/to/data

where path/to/data contains the CHUBS dataset (extracted from the glyphs.zip zip file).

For more options, see args.py or execute python -h train_vit.py or python -h train_resnet.py.