Skip to main content

Visual Embedder

Learn how to train a visual embedding model using Clarifai SDKs

Visual embedder models are neural network architectures specifically designed to transform high-dimensional visual data, such as images or videos, into low-dimensional representations, called embeddings. You can learn more about Visual Embedder here.

App Creation

The first part of model training includes the creation of an app under which the training process takes place. Here we are creating an app with the app id as “demo_train” and the base workflow is set as “Universal”. You can change the base workflows to Empty, Universal, Language Understanding, and General according to your use case.

from clarifai.client.user import User
#replace your "user_id"
client = User(user_id="user_id")
app = client.create_app(app_id="demo_train", base_workflow="Universal")

Dataset Upload

The next step involves dataset upload. You can upload the dataset to your app so that the model accepts the data directly from the platform. The data used for training in this tutorial is available in the examples repository you have cloned.

#importing load_module_dataloader for calling the dataloader object in in the local data folder
from clarifai.datasets.upload.utils import load_module_dataloader

# Construct the path to the dataset folder
module_path = os.path.join(os.getcwd().split('/models/model_train')[0],'datasets/upload/image_classification/food-101')

# Load the dataloader module using the provided function from your module
food101_dataloader = load_module_dataloader(module_path)

# Create a Clarifai dataset with the specified dataset_id
dataset = app.create_dataset(dataset_id="image_dataset")

# Upload the dataset using the provided dataloader and get the upload status

If you have followed the steps correctly you should receive an output that looks like this,


Choose The Model Type

First let's list the all available trainable model types in the platform,


Click here to know more about Clarifai Model Types.

Model Creation

From the above list of model types we are going to choose visual-embedder as it is similar to our use case. Now let's create a model with the above model type.

MODEL_ID = "model_visual_embedder"
MODEL_TYPE_ID = "visual-embedder"

# Create a model by passing the model name and model type as parameter
model = app.create_model(model_id=MODEL_ID, model_type_id=MODEL_TYPE_ID)

Template Selection

Inside the Clarifiai platform there is a template feature. Templates give you the control to choose the specific architecture used by your neural network, as well as define a set of hyperparameters you can use to fine-tune the way your model learns. We are going to choose the 'Clarifai_ResNext' template for training our model.


Setup Model Parameters

You can update the model params to your need before initiating training.

# Get the params for the selected template
model_params = model.get_params(template='Clarifai_ResNext')
# list the concepts to add in the params
concepts = [ for concept in app.list_concepts()]
model.update_params(dataset_id = 'image_dataset',concepts = concepts)
{'dataset_id': 'image_dataset',
'dataset_version_id': '',
'concepts': ['id-hamburger', 'id-ramen', 'id-prime_rib', 'id-beignets'],
'train_params': {'invalid_data_tolerance_percent': 5.0,
'template': 'Clarifai_ResNext',
'logreg': 1.0,
'image_size': 256.0,
'batch_size': 64.0,
'init_epochs': 25.0,
'step_epochs': 7.0,
'num_epochs': 65.0,
'per_item_lrate': 7.8125e-05,
'num_items_per_epoch': 0.0}}

Initiate Model Training

We can initiate the model training by calling the model.train() method. The Clarifai SDKs also offers features like showing training status and saving training logs in a local file.


If the status code is 'MODEL-TRAINED', then the user can know the Model is Trained and ready to use.

import time
#Starting the training
model_version_id = model.train()

#Checking the status of training
while True:
status = model.training_status(version_id=model_version_id,training_logs=False)
if status.code == 21106: #MODEL_TRAINING_FAILED
elif status.code == 21100: #MODEL_TRAINED
print("Current Status:",status)

Model Prediction

Since the model is trained and ready let’s run some predictions to view the model performance,

IMAGE_PATH = os.path.join(os.getcwd().split('/models')[0],'datasets/upload/image_classification/food-101/images/hamburger/139558.jpg')
model_prediction = model.predict_by_filepath(IMAGE_PATH, input_type="image")

# Get the output
embeddings {

vector: 0.021010370925068855

vector: 0.011909130029380322

vector: 2.2577569325221702e-07

vector: 0.001307532424107194

vector: 0.04247743636369705

vector: 0.01022490207105875

vector: 0.0006444881437346339

vector: 0.027988344430923462

vector: 0.028407510370016098

vector: 5.129506917000981e-06

vector: 0.03279731422662735

vector: 0.016899824142456055

vector: 0.003125722287222743

vector: 0.0

vector: 0.024156155064702034

vector: 0.04975743591785431

vector: 0.010608416981995106

vector: 0.0006941271130926907

vector: 0.00018513976829126477

vector: 2.714529364311602e-05

vector: 0.0014789806446060538