Predicting the Past: Estimating Historical Appraisals with OCR and Machine Learning

Paper: arxiv Dataset: huggingface

Setup

Start local PostgreSQL server
Create GoogleDrive credentials (follow instructions steps 1-5 using your personal Google Account)
Populate database using fill_db.py

Name		Name	Last commit message	Last commit date
Latest commit History 128 Commits
Dataset/Ownership		Dataset/Ownership
clean_data		clean_data
documents		documents
explore_data		explore_data
figures		figures
helpers		helpers
label_data		label_data
load_data		load_data
modeling_pipeline		modeling_pipeline
ocr_pipeline		ocr_pipeline
.gitignore		.gitignore
Readme.md		Readme.md
bias_analysis.ipynb		bias_analysis.ipynb
create_testset.py		create_testset.py
crop_images.py		crop_images.py
current_best.log		current_best.log
image_processing_experiments.ipynb		image_processing_experiments.ipynb
ocr.py		ocr.py
ocr_filtering.py		ocr_filtering.py
ocr_targets.sql		ocr_targets.sql
paper_figures.ipynb		paper_figures.ipynb
push_huggingface_data.py		push_huggingface_data.py
requirements.txt		requirements.txt
segmentation_experiments.ipynb		segmentation_experiments.ipynb
training.log		training.log
triage.log		triage.log
wandb_data_growth_sweep.csv		wandb_data_growth_sweep.csv