gaborpelesz / arbori

not maintained image classification repo

AR bori python image classification backend (a.k.a. Black Box API)

Init/Install project

First clone the repo.

git clone https://${USERNAME}@bitbucket.org/exahexa/ar-bori-model.git

At first, the command make dev will create the docker image (After that you can rebuild with make build)

make run

Makefile commands

# Build the python service image and force remove previous instance
make build

# Push to docker hub
make push

# Run all the service containers with compose (docker-compose up) and force recreate container for python service
make run

# Run developement environment (make run with docker-compose.dev.yml)
make dev

# Get into the python service's container shell
make exec

# Turn down running docker-compose services
make down

---

Usage

Server will be listening on port 3000

API endpoints:

• POST /predict

POST /predict

Receives an image and tries to predict its class.

• request: POST /predict
• required headers: {Content-Type: application/json}

JSON properties:

{
  "image": "/9j/4ROSR ... FAf/Z"
}

• image: Base64 encoded image that contains the object.

Response:

{
    "status": "Success", 
    "prediction": {
        "class": "phone", 
        "probability": "0.9996686"
        }, 
    "processing_runtime": "1125.247ms"
}

• status: Indicates the result of the request.
• prediction:
  • class: The predicted class of the image.
  • probability: The probability of the image being the instance of that class
• processing_runtime: The total runtime of transforming and predicting the image's class

---

Error handling (HTTP 400, HTTP 500)

All internal server errors and bad request has the following JSON response structure sent with the appropriate status code:

{
  "status": "Failed",
  "status_message": " ...Developers explonation... ",
  "error_message": " ...Error from the application... "
}

---

Postman

Postman collection file: postman_example/arbori.postman_collection.json.

© gaborpelesz 2019