seeme 0.21.1

No-Code/Low-code MLOps: Create, Operate, and integrate machine learning models in a standardized way.

Welcome to SeeMe.ai

SeeMe.ai is a no/low code MLOps platform aiming to be the simplest way you create, use, and share AI.

You can use SeeMe.ai without any code to automate the full AI lifecycle of your datasets and models.

The Python SDK gives easy access to all of your datasets, models, jobs, ... on the platform.

Installation

$ pip install seeme

Getting started

from seeme import Client

cl = Client()

# -- Registration --
my_username =    # example: "my_username"
my_email =       # example: "[e-mail]"
my_password =    # example: "supersecurepassword"
my_firstname =   # example: "Jan"
my_name =        # example: "Van de Poel"

cl.register(
    username=my_username, 
    email=my_email, 
    password=my_password, 
    firstname=my_firstname, 
    name=my_name
)

# -- Log in --
cl.login(username, password)

# -- Log out --
cl.logout()

Datasets

Manage the entire lifecyle of your datasets:

• create
• manage
• version
• label
• annotate
• import/export

from seeme import DATASET_CONTENT_TYPE_IMAGES

# -- Get datasets --
datasets = cl.get_datasets()

my_dataset = {
    "name": "Cloud classifier",
    "description": "Classify clouds from pictures",
    "default_splits": True, # If `True`, adds 'train', 'valid', and 'test' default_splits
    "content_type": DATASET_CONTENT_TYPE_IMAGES,
    "multi_label": False
}

my_dataset = cl.create_dataset(my_dataset)

Checkout the dataset documentation to see all possibilities and detailed guides.

Models

Manage the entire lifecycle of your AI models:

• create
• managem
• version
• converst
• predicti
• import/export

# -- Get models --
models = cl.get_models()

# -- Application ID --
application_id = cl.get_application_id(
    base_framework="pytorch",
    framework="fastai",
    base_framework_version=str(torch.__version__),
    framework_version=str(fastai.__version__),
    application="image_classification"
)

# -- Create model --
model_name = "Cloud classifier"
description = "Classify clouds from images"

my_model = cl.create_model(  {
    "name": model_name,
    "description": description,
    "application_id": application_id,
    "auto_convert": True # Automatically converts your model to ONNX, CoreML, and TensorFlow Lite.
})

cl.upload_model(my_model["id"], model_file_location)

cl.inference(my_model["id"], image_location)

Checkout the [model] documentation](https://docs.seeme.ai/python-sdk/#models) to see all possibilities and detailed guides.

Jobs

Schedule training, validation, and model conversion jobs with a simple command:

from seeme import JOB_TYPE_TRAINING

jobs = cl.get_jobs()

job = {
    "name": "v3 image classifier",
    "description": "A new dataset for an improved model",
    "status_message": "",
    "application_id": application_id,
    "job_type": JOB_TYPE_TRAINING,
    "dataset_id": dataset_id,
    "dataset_version_id": datset_version_id,
    "model_id": model_id,
    "model_version_id": model_version_id,
    "items": [
      {
        "name": "image_size",
        "value": "224",
        "value_type": "number"
      },
      {
        "name": "arch",
        "value": "resnet50",
        "value_type": "text"
      }
    ]
  }

Applications

SeeMe.ai automates the full lifecycle of data and models for a wide range of AI applications, such as:

• image classification
• object detection
• structured data
• language models
• multi lingual text classification
• object character recognition (OCR)
• named entity recognition (NER)

for a number of AI frameworks and their versions:

• fastai
• ONNX
• Core ML
• TensorFlow Lite
• PyTorch
• Yolo v4
• Spacy
• Tesseract
• PaddleOCR

For a full list of frameworks and their versions:

# -- Get applications --
all_applications = cl.get_applications()

print(all_applications)

SDK Documentation

For more detailed SDK documentation see https://docs.seeme.ai.