README.md

---
title: Small Object Detection
emoji: 🔍
sdk: docker
app_port: 7860
pinned: false
---

# Small Object Detection

Upload an image to detect objects using the trained YOLO model **`best.pt`** in this repo. **CPU-only** — runs on basic (free) Hugging Face Spaces. The `train26m` folder is not part of this repo; only `best.pt` is included.

## Run locally

**Using uv (recommended):**
```bash
pip install uv
uv pip install -r requirements.txt
python app.py
```

**Or with pip only:**
```bash
pip install -r requirements.txt
python app.py
```

Then open the URL shown in the terminal (e.g. http://127.0.0.1:7860).

## Docker (Space)

The Space builds from the Dockerfile using **Python 3.10** and **requirements-lock.txt** so the container matches a known set of versions. To match your local env exactly, from your venv run:

```bash
pip freeze > requirements-lock.txt
```

Then commit and push; the next Space build will use those exact versions.