A Vision Transformer specialised in reading and recording receipts using MiniCPM-V-4_5-int4. Automatically processes all images in a folder, analyzes them, and generates a CSV file for easy auditing tasks.
GPU Highly Recommended.
git clone https://github.com/walterkwong/vit_receipt
cd vit_receipt
pip install .openbmb/MiniCPM-V-4_5-int4
-
Create an access token on Hugging Face.
-
Authenticate with your token:
hf auth login
Note:
The first run will download the model (~6.54 GB). Ensure you have sufficient disk space and a stable internet connection. Also, this is a very immature project solely for personal use, but do let me know if you want me to work on this project.
Model used: openbmb/MiniCPM-V-4_5-int4