python-redactor

Description

Running server.py will start a Flask server that accepts POST requests containing a file in jpg, jpeg, png formats at localhost:5000.

It will return a JSON containing "boxes", an array of multiple arrays each of which contains a starting X coordinate, starting Y coordinate, ending X, and ending Y that corresponds to a bounding box around any text the EAST detection model sees.

The JSON also contains a "text" array that contains strings corresponding to the same index as the box it describes in the "boxes" array.

Requirements

• Tesseract
• Python Packages
  • pytesseract
  • opencv-python
  • Flask
  • numpy

To install the python packages, run

pip install <package name>

If you choose to build an image with Docker, then this step is unnecessary as they will be built together.

Usage With Docker

Tested using the Postman Desktop App

1. Make sure Docker is installed on your machine
2. Run

docker pull jwn8175/xray-redactor

3. After the image is done pulling, run

docker run -p 5000:5000 jwn8175/xray-redactor

which should show the IP the server is hosted on

Postman Method

4. In the Postman Desktop App, setup a new request to the IP above
5. Change the request type to POST
6. Click body in the navigation bar of that request, and click "form-data".
7. Include a file with the key "file" that is a picture with text of the form jpg, jpeg, or png. It should look something like the picture below
8. Send the request and you should receive a JSON of the aforementioned format described in the description. An example is below

{
  "boxes": [
    [
      364,
      15,
      505,
      46
    ],
    [
      1,
      16,
      183,
      62
    ],
    [
      358,
      462,
      519,
      496
    ],
    [
      2,
      415,
      107,
      495
    ]
  ],
  "text": [
    "NAME NAME\nNAME MEDICAL SYSTEMS.\n\f",
    "NAME, M, 129838\n927, 721389\nFr: 4, WL: 40, WW: 400\n\f",
    "1091 ms\nOFOV: 360.000000 mm\n\f",
    "LightSpeedUltra,\nkv: 120\n\nmA: 40\n\n5.000000 mm\nTilt: 0.000000\n\f"
  ]
}

curl Method

4. CD to the directory that holds the file you wish to send to the program
5. Send a command in the form

curl -X POST -F "file=@<filename>" localhost:5000

Without Docker

If you choose not to use Docker, install the required packages manually, pull from the repository, run server.py, and continue from Step 3 above.

Building Docker Container

After pulling from the repository, cd to the directory and run

docker build -t python-redactor .

To use the container when it is done building, follow from step 3 in the usage steps above.