This repository contains an implementation of YOLO-NAS (You Only Look Once - Neural Architecture Search) for 3D computer vision tasks. YOLO-NAS represents a significant advancement in object detection, combining the speed of YOLO models with neural architecture search to achieve state-of-the-art performance.
- High-Performance Object Detection: Leveraging YOLO-NAS architecture for efficient and accurate object detection
- 3D Vision Capabilities: Extended for processing and analyzing 3D spatial data
- Optimized Architecture: Neural architecture search for optimal model performance
- Easy Integration: Simple API for training and inference
- Customizable: Easily adaptable to various 3D vision tasks
- Python 3.8+
- PyTorch 1.10+
- CUDA 11.3+ (for GPU acceleration)
- Other dependencies listed in
requirements.txt
-
Clone the repository:
git clone https://github.com/tarak6984/3d-computer-vision.git cd 3d-computer-vision -
Install dependencies:
pip install -r requirements.txt
To train the model on your custom dataset:
python train.py --config yolonas.ymlTo run inference on images or video:
python inference.py --source path/to/input --output path/to/outputTo evaluate model performance:
python evaluation.py --weights path/to/weights --data path/to/dataset| Model | mAP@0.5 | FPS (GPU) | Parameters (M) |
|---|---|---|---|
| YOLO-NAS-S | 47.5 | 42 | 12.8 |
| YOLO-NAS-M | 51.5 | 33 | 23.0 |
| YOLO-NAS-L | 53.0 | 25 | 36.5 |
3d-computer-vision/
├── README.md # Project documentation
├── requirements.txt # Python dependencies
├── train.py # Training script
├── inference.py # Inference script
├── evaluation.py # Model evaluation
├── params.py # Model parameters
├── yolonas.yml # Configuration file
└── readme.png # Example output
Contributions are welcome! Please feel free to submit a Pull Request.
This project is licensed under the MIT License - see the LICENSE file for details.
For any questions or suggestions, please open an issue or contact tareksabbir20@gmail.com.