README.md

---
title: "NOVIC: Unconstrained Open Vocabulary Image Classification"
short_description: "Prompt-free open vocabulary classification of any image"
license: gpl-3.0
emoji: 🖼️
colorFrom: green
colorTo: yellow
sdk: gradio
python_version: "3.10"
sdk_version: "5.35.0"
app_file: app.py
fullWidth: true
header: default
tags:
  - prompt-free
  - open vocabulary
  - image classification
  - computer vision
  - identification
  - generative
pinned: true
disable_embedding: false
---

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

To run the Gradio space locally, you first create a sandboxed Python 3.10 environment:
```bash
python -m venv .venv  # <-- Python 3.10
source .venv/bin/activate
# OR...
conda create -y -n novic_spaces python=3.10
conda activate novic_spaces
```
Then install `gradio` and any further requirements (**Note:** A pip-based install of `torch` is actually numerically different than a conda-based install of `pytorch`, which matters for the conda-based pretrained models, but for Hugging Face spaces we do not really have another choice):
```bash
pip install gradio==5.35.0 -r requirements.txt
```
Now run the Gradio application:
```bash
./app.py
# OR...
gradio app.py
```
Open in a browser the URL that the application is running on (as shown in the CLI output), e.g. [http://127.0.0.1:7860](http://127.0.0.1:7860).