AxelSariel / diy-ml

DIY ML API Service for EC530

diy-ml

DIY ML API Service for EC530

DIY ML is an API to train, test, and deploy ML models.

Future implementation will include actual ML training using https://auto.gluon.ai/stable/index.html

Screenshots

User Stories

Below is a list of User Stories for reference, from the class slides

• API user should be able to create a ML image classification or Object detection project for Training and Inference
• API user should be able to upload data (images) for training in a project
• API user should be able to upload label or class data for images in a project
• API user should be able to analyze data before training
• API user should be able to add or remove training points
• API user should be able to configure training parameters
• API user should be able when the training is completed to get training stats
• API user should be able to test a model using new dataset and get results
• API user should be able to deploy a model to be used for inference and should be able to get a unique API to use for a project-iteration combination
• API user should be able to run and track iterations of training
• API user should be able to use inference API to run and get results on an image
• ALL APIs should be independent of the ML model and data
• A project is associated with a user

Routes (some TODO)

Endpoint Methods Rule

---

datasets_create POST /diyml/datasets/create
datasets_objects_create POST /diyml/datasets/objects/create
datasets_objects_delete POST /diyml/datasets/objects/delete
index GET /
inference_delete POST /diyml/inference/delete
inference_deploy POST /diyml/inference/deploy
inference_infer POST /diyml/inference//infer preprocess POST /diyml/preprocess
projects_create POST /diyml/projects/create
projects_delete POST /diyml/projects/delete
test_create POST /diyml/test/start
test_results GET /diyml/test/results
test_status POST /diyml/test/status
test_stop POST /diyml/test/stop
training_results GET /diyml/training/results
training_start POST /diyml/training/start
training_status GET /diyml/training/status
training_stop POST /diyml/training/stop
users_create POST /diyml/users/create

Setup

1. Clone repo

git clone https://github.com/AxelSariel/diy-ml
cd diy-ml

2. Setup Python Virtual Environment

python3 -m venv ml

3. Activate Virtual Environment

source ml/bin/activate

4. Install Requirements

pip install -r requirements.txt

5. Run Server

flask run

Docker Run

1. Build the docker image

docker build -t diyml .

2. Run the docker container

docker rum diyml

Export Docker Image

1. Export image

docker save -o diyml.tar diyml

2. Import image on another computer

docker load -i diyml.tar