Timeline #2
Description
Mid-March:
Multilingual initial results @leejejang
- Train and test TESTR-v3 on four language groups
Historical map clustering and Synthetic map generation
- Generate synthetic maps with different styles @zekun-li
- Verify if TESTR-v3 model will get better performance with synthetic data
- Provide map-name list for each cluster @zekun-li
- Fine-tune TESTR-v3 @leejejang
Integration @tanisha
- Finish integration, with both Recogito and mapKurator-system located within the docker, supports three types of inputs: 1) local file 2) IIIF 3) WMTS
==================================================================
Mid-April:
Multilingual @leeje
- Deliver results for the whole map collection with V2
- Integrate with mapKurator-system pipeline, support easy switch between different models (testr ,v2, v2-multi-languages)
- Documentation about results
Historical map clustering and Synthetic map generation
- Verify if TESTR-v3 model will get better performance with synthetic data
- Evaluate Fine-tuned TESTR-v3 @leejejang
Sanborn & Ordnance Survey maps @leejejang
- Run v2, deliver results
Integration: @tanisha
- Update docker image
- Documentation
- Doc sanity check, follow the steps for installation. @leejejang