Art Segmentation

Download weights from Google Drive

    gdown --id 1EDUARyROWS1yUbqTcZ0AhJ7K-31M9MZk
    unzip weight.zip

Manually from bash https://drive.google.com/file/d/1EDUARyROWS1yUbqTcZ0AhJ7K-31M9MZk/view?usp=drive_link

Install Required Dependencies

Create Virtual Environment

    conda create --name artdetector python=3.8
    conda activate artdetector

Install Dependencies

    pip install -r requirements.txt

Run Flask program

    python application.py

Run Browser to test segmentation

 http://127.0.0.1:5000

Choose your image and click the Upload button
Wait until you get the segmented output

Click the ReTake Button to test new images

API

URL: http://127.0.0.1:5000/detectart

OPEN POSTMAN: POST body

    {
        "image":"<base64 image>"
    }

Run Docker

    docker build -t <your_tag> .
    docker container run -p 5001:5000 <your_tag>

Deploying Art Detector AI Engine in Elastic BeanStalk

Make .zip folder with following hierachy

Flask app.
│   application.py
│   src
│   README.md
│   requirements.txt
│
├───.ebextensions
│       packages.config
│       python.config
│
├───.platform
│   └───nginx
│       └───conf.d
│               upload_size.conf
│
├───templates
│       index.html
│       output.html
|───weights

NOTE: The name of your main flask app must be application.py

Install required packages in Elastic BeanStalk Environment

1. Create a folder name .ebextensions
2. Create file packages.config name and paste the following lines

commands:
    install_mesa:
        command: yum install mesa-libGL -y

The above command install libGL library which is required for OpenCV and rembg

3. Create file python.config

option_settings:
  "aws:elasticbeanstalk:container:python":
    WSGIPath: application:application

4. Create folder name .platform/nginx/conf.d and file .platform/nginx/conf.d/upload_size.conf

client_max_body_size 20M;
proxy_read_timeout 3000;
proxy_connect_timeout 3000;
proxy_send_timeout 3000;

client_max_body_size: proxy_read_timeout: proxy_connect_timeout: proxy_send_timeout:

Create Elastic BeanStalk Application

1. Open Elastic BeanStalk
2. Click Create environment Fill Following Information:

Application Information Application name: Artdetector

Platform platform: Python platform branch: Python 3.8 platform version: 3.5.3

Application code Upload your code Version label: ver-1 Local file: Upload application (Choose File)

Presents: Single instance

Click Next

3. Service role:
   • Create and use new service role

EC2 key pair: - Select your keys

EC2 instance profile: -

Create role with name: aws-elasticbeanstalk-ec2-role Goto: IAM > Roles > Create role Add Permission details:

    AWSElasticBeanstalkEnhancedHealth
    AdministratorAccess-AWSElasticBeanstalk
    AWSElasticBeanstalkManagedUpdatesCustomerRolePolicy

Click Next

4. Configure instance traffic and scaling

Instance settings: Public IP address: tick: Activated

Click Next

Add EC2 Security groups: Enable Port 5000 - 5005

Auto scaling group: Select your environment type: Single Instance or Load Balance

Select instance t3.x2large, t3.xlarge

Click Next

5. Configure updates, monitoring, and logging Rolling updates and deployments: Command timeout: 3500

Click Next

6. Submit