This is a recipet scanner based on registering of image and then doing ocr to extarct the numerical values
In future I might update it to extarct all the informations like items description, date and all the information and redirect it into csv
The result came out to be 14 as the total amount
I have also added functionality of making a PDF chatbot for invoice scanning which takes PDF as an input parse it and then I made the vector database using FAISS.
The LLM that was used is LLAMAv2-7B-4bit(you can also infer it on google colab)
The prompt feeding, inferencing, chaining was done by Langchain
The prompt is set such that the output should be precise only