API.md

Continuous Integration and Continuous Delivery

Author: Pradyumna Krishna
UPDATED: 28/03/2021

Today, I made an API which predicts emotion, this api can be used publicly or by anyone in any project. This API returns response with no of faces, their coordiantes, emotions and probability of the guessed emotion. This API uses my TensorFlow model which has 70% testing and 70% validation accuracy.

How to use

The API url is /FER/api/v1/ present at the latest website hosted on (possible that website is no longer hosted), send a POST request using curl or postman, and uploading the file.

CURL Request Example

curl -F image=@<path-to-file> http://<domain>/FER/api/v1/
# a running example
curl -F [email protected] http://127.0.0.1:5000/FER/api/v1/

Using Python requests

import requests
url = 'http://127.0.0.1:5000/FER/api/v1/
files = {'image': open('face.jpg', 'rb')}
requests.post(url, files=files)

It will give you response accordingly, Example JSON response is present below:

{
    "data": {
        "0": {
            "coordinates": {
                "h": 284,
                "w": 284,
                "x": 397,
                "y": 138
            },
            "emotion": "Neutral",
            "probability": {
                "Angry": 0.05442037433385849,
                "Disgust": 0.0001229508634423837,
                "Fear": 0.023583699017763138,
                "Happy": 0.0007338053546845913,
                "Neutral": 0.7785017490386963,
                "Sad": 0.14203275740146637,
                "Surprise": 0.0006046505877748132
            }
        }
    },
    "message": "Yes! 1 face(s) detected!"
}

• Point to Note: the image is resized to max-width of 1200px and max-height of 600px. Thats why, the response coordinates are also changed.

Back: Setup
Next: The Web Application