Harness the power of text to add computer vision to your app in minutes

Zeroshot is an open source tool for developers that creates image classifiers from text descriptions

Open Source

Zeroshot is a free, community-driven product for versatile applications.

Low Latency

Fast and efficient real-time image classification. More accurate and faster than OpenAI's CLIP.

Offline Access

No internet required, you can use Zeroshot and deploy from anywhere, anytime.

Build a classifier in seconds, with no labeling required.

Deploy to CPU or GPU easily with a few lines of code

    import zeroshot

    # Create the classifier and preprocessing function.
    classifier = zeroshot.Classifier(
        "your model string or path"
    preprocess_fn = zeroshot.create_preprocess_fn()

    # Run the model! The result is the index of the prediction.
    prediction = classifier.predict(preprocess_fn(image))
    print(f"The image is class {prediction}")

Save costs with a fast, local model that doesn't cost money each time it's run.

How it works

Train your model

Train your custom model in only a few minutes with our simple web interface. No data required.

Install Zeroshot

Install Zeroshot with pip or conda. Zeroshot is open source and free to use.

Import Zeroshot

Import Zeroshot into your project and use it to classify images.

Use with your custom model

To start using your custom trained model, just paste in the UUID generated from the web interface.