brain-facens/fieldvision-ai

API for processing text on invoices, with the aim of identifying relevant fields on an invoice and optimizing bonus or validation systems.

FieldVision AI

API_demo.mp4

API for processing text on invoices, with the aim of identifying relevant fields on an invoice and optimizing bonus or validation systems. Making life easier for logisticians, merchants and managers, the application has an interface that captures images from the webcam, processes the image using OCR and provides a visualization of the results obtained.

Docs

Visit our wiki

First Steps

This project was developed to run a web interface, where the user will have access to the image capture system and the data processed from that image. Follow the steps below to use this project, the stable version is in the main branch.

Requirements

The following requirements should ideally be met for proper operation:

• Ubuntu 20.04 (Original development environment, but compatible with 18.04 and 22.04)

• Docker Engine or Docker Desktop

• Python 3.8

• Git

• Anaconda/Miniconda

• Create a virtual environment for the project if you want to work without conda;

  conda create --name fielvision python=3.8 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
  
  # Activate the environment
  conda activate fielvision
  

---

Warning! We recommend that you use the Docker container developed for this project, as the application is stable. This avoids the risk of incorrect installations, path errors and library versions.

---

Installation

Local:

# Activate the environment
conda activate fielvision

# Clone repository
git clone https://github.com/brain-facens/FieldVision-AI.git

# Install requirements
cd FieldVision-AI/
pip install -r requirements.txt

Docker:

# Pull image
docker pull brain20/ocr-notas

Usage

Local:

# Applicable filter of up to 3 words 
# python src/field_vision_API/main.py <fist, second, third>

# Run API
python src/field_vision_API/main.py

Docker:

# Running the container with the application
docker run -it --rm -p 8080:8080 brain20/ocr-notas

---

Warning! Running the container will start the API, which is the interface between OCR processing and the end user. The API is documented in a Swagger, where you can test it.

---

Demo

API_demo_2.mp4

---

🤝 Collaborators

We would like to thank the following people who contributed to this project:

Natanael Vitorino

Lucas Oliveira

Pedro Gabriel

📝 License

This project is under license. See the file LICENSE for more details.

---

fieldvision-ai's Issues

Documentação

Realizar a atualização da documentação do código e aplicação de padrões de qualidade de software.

• Atualizar o README
• Docstrings
• Padrão de código
• Review

The documents are not up to date and have some errors

Describe the bug
When I was reading it, I found some small details wrong, for example: the name of the repository has been updated, but not in the documentation, as you can see in the image below.

Auto tests on API endpoints

• API deploy
• Automed tests on endpoints

Validação da inserção de string para filtragem

@pedroogaab e @LucasLCO Ao testar o código, me deparei com um erro de implementação por usuário. Este erro é simplismente enviar a imagem pelo método POST e não declarar nenhuma palavra como filtro pelo método PUT, resultando assim no erro de objeto não interativo. Abaixo demonstro os passos para replicar o bug.

POST da imagem para processamento

PUT do filtro

Console da API