Poor Man’s Rekognition (PMR) is an open-source alternative to amazon’s Rekogniton API service. It is capable of facial analysis, object detection, text detection, scene detection and has the potential to be much more.

What has been added during this summer?

This year the project has focused on being more compatible for developers, providing a snappier experience for the end-users, and making a robust CI for maintainers.

Major changes are listed as follows:

Before GSoC:

  1. Fixing scene detection model->LINK
  2. Fixing error logs -> LINK

During GSoC:

  1. Dev environment compatible with python3.6–3.8-> LINK
  2. LANMS pip package for python3.6–3.8-> LINK
  3. Fixing CI pipeline-> LINK
  4. Adding tests for video features-> LINK
  5. Added offline testing script -> LINK
  6. Upgrading the project to TensorFlow V2 ->LINK
  7. Adding TextBoxes++ for SSD textbox detection-> LINK
  8. Adding efficientdet for object-detection with 12 new classes-> LINK
  9. Making a docker file for containerized deployment-> LINK

To Be Done:

  1. Adding Image Captioning model
  2. upgrading scene detection model to efficientnet

How can you use the service?

Now for the end-user, you might want to use a fully abstracted version of PMR so that you can integrate it with your application. In that case, I would recommend you to go to the dockerfile way.

step 1: Clone the repository

git clone

Step 2: Build a docker image

docker image build ./

this will print a unique image ID at the end. Note that down.

Step 3:

run the built image with port 8000 exposed

docker run -p 8000:8000 <IMAGEID>

Now you are all set to make API calls to PMR.

Watch this video to see how->


For Developers, please refer to for setting up the project in a development environment.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store