# Vehicle Classification — UTD Deep Learning Assessment

A Convolutional Neural Network (CNN) trained to classify 8 vehicle types:
**Bicycle, Bus, Car, Motorcycle, NonVehicles, Taxi, Truck, Van**

---

## Requirements
- Python 3.11 (see `.python-version`)
- pip

---

## Setup

### 1. Clone the repository
```bash
git clone https://git.keshavanand.net/KeshavAnandCode/vehicle-classification.git
cd utd-vehicle-classification
```

### 2. Create and activate a virtual environment
```bash
python3.11 -m venv .venv
```
**Mac/Linux:**
```bash
source .venv/bin/activate
```
**Windows:**
```bash
.venv\Scripts\activate
```

### 3. Install dependencies
```bash
pip install -r requirements.txt
```

### 4. Download the dataset
- Download `vehicle_classification.zip`
- Place and extract it so the structure looks exactly like this:
```
data/
└── raw/
    └── vehicle_classification/
        ├── Bicycle/
        ├── Bus/
        ├── Car/
        ├── Motorcycle/
        ├── NonVehicles/
        ├── Taxi/
        ├── Truck/
        └── Van/
```

### 5. Run the notebook
```bash
jupyter notebook notebooks/submission.ipynb
```
Once open: **Kernel → Restart & Run All**

---

## Project Structure
```
utd-vehicle-classification/
├── notebooks/
│   ├── submission.ipynb          ← main submission, run this
│   └── experiments/              ← exploratory notebooks (ignore)
├── data/
│   └── raw/
│       └── vehicle_classification/   ← dataset goes here
├── models/                       ← saved model weights (auto-created on run)
├── results/                      ← training curves (auto-created on run)
├── requirements.txt
└── README.md
```

---

## Expected Output
```
Final Train Accuracy : ~88%
Final Test  Accuracy : ~82%
```

---

## Notes
- Device is detected automatically — runs on NVIDIA GPU, Apple Silicon (MPS), or CPU with no code changes required
- `models/` and `results/` directories are created automatically when the notebook runs