Portrait Segmentation using Tensorflow

This script removes the background from an input image. You can read more about segmentation here

Setup

The script setup.sh downloads the trained model and sets it up so that the seg.py script can understand.

./setup.sh

Running the script

Go ahead and use the script as specified below, to execute fast but lower accuracy model:

python3 seg.py sample.jpg sample.png

For better accuracy, albiet a slower approach, go ahead and try :

python3 seg.py sample.jpg sample.png 1

Dependencies

tensorflow, PIL

Sample Result

Input:

Output:

Docker for API

You can build and run the docker using the following process:

Cloning

git clone https://github.com/jqueguiner/background-removal.git background-removal

Building Docker

cd background-removal && docker build -t background-removal -f Dockerfile-api .

Running Docker

echo "http://$(curl ifconfig.io):5000" && docker run -p 5000:5000 -d background-removal

Calling the API for image detection

curl -X POST "http://MY_SUPER_API_IP:5000/process" -H "accept: image/png" -H "Content-Type: application/json" -d '{"url":"https://i.ibb.co/W0JpjrY/input.jpg"}' --output no-background-image.png

Image Background Removal API will remove the background of the provided image using Deep Learning and especially image segmentation.

read this paper to learn more about this technic: Removing Background with Semantic Segmentation Based on Ensemble Learning

@INPROCEEDINGS{10.4108/eai.21-6-2018.2276586,
    author={Junhong Xu and Hanqing Guo and Aron Kageza and Shaoen Wu and Saeed AlQarni},
    title={Removing Background with Semantic Segmentation Based on Ensemble Learning},
    proceedings={11th EAI International Conference on Mobile Multimedia Communications},
    publisher={EAI},
    proceedings_a={MOBIMEDIA},
    year={2018},
    month={9},
    keywords={deep learning image segmentation background removal},
    doi={10.4108/eai.21-6-2018.2276586}
}

EXAMPLE

INPUT

{
  "url": "https://i.ibb.co/W0JpjrY/input.jpg"
}

EXECUTION FOR DISTANT FILE (URL)

curl -X POST "https://api-market-place.ai.ovh.net/image-background-removal/process" -H "accept: image/png" -H "X-OVH-Api-Key: XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX" -H "Content-Type: application/json" -d '{"url":"https://i.ibb.co/W0JpjrY/input.jpg"}' --output no-background-image.png

EXECUTION FOR LOCAL FILE (UPLOAD)

curl -X POST "https://api-market-place.ai.ovh.net/image-background-removal/process" -F file=@input.jpg -H "accept: image/png" -H "X-OVH-Api-Key: XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX" --output no-background-image.png

OUTPUT Binary output image.

please refer to swagger documentation for further technical details: swagger documentation