ourcustompkg 0.2.1

A robust YOLOv7-based package designed for efficient toy car detection and comprehensive dataset management.

OurCustomPkg: YOLOv7-based Toy Car Detection

Welcome to OurCustomPkg, a cutting-edge Python package designed for the detection of toy cars using the powerful YOLOv7 model. Additionally, the package integrates hand-tracking capabilities via MediaPipe, allowing for interactive and dynamic detection experiences.

Features

• YOLOv7 Integration: Utilize the state-of-the-art YOLOv7 model for accurate and efficient toy car detection.
• Multi-Source Input: Supports images, video files, and real-time webcam feeds as input sources.
• Hand Tracking: Employ MediaPipe's hand tracking to interact with detected objects in real-time.
• Highly Customizable: Easily adjust detection parameters and extend functionalities according to your project needs.

Installation

To get started with OurCustomPkg, you can install it directly from PyPI:

pip install ourcustompkg

This command will install the package along with all the necessary dependencies, including PyTorch, OpenCV, and MediaPipe.

Getting Started

Basic Usage

The primary script for toy car detection is detect_car.py, located in the ourcustompkg/yolov7/ directory. Here's how to use it:

python -m ourcustompkg.yolov7.detect_car --source <input_source> --weights <path_to_weights>

Example Commands

Detect cars in an image:

python -m ourcustompkg.yolov7.detect_car --source data/images/car.jpg --weights yolov7.pt

Detect cars from a video file:

python -m ourcustompkg.yolov7.detect_car --source data/videos/car_video.mp4 --weights yolov7.pt

Real-time detection using a webcam:

python -m ourcustompkg.yolov7.detect_car --source 0 --weights yolov7.pt

Hand Tracking Interaction

One of the standout features of OurCustomPkg is its integration of hand-tracking functionality using MediaPipe. When running the detect_car.py script, you can interact with detected cars using hand gestures tracked by your webcam.

Command-Line Arguments

• --source: Specifies the input source, which can be an image file, video file, or webcam feed.
• --weights: Path to the YOLOv7 weights file. You can download pretrained weights from the official YOLOv7 repository.
• --img-size: Sets the size of the input image for detection (default: 640).
• --conf-thres: Confidence threshold for filtering weak detections (default: 0.25).
• --iou-thres: IoU threshold for non-maximum suppression (default: 0.45).
• --device: Specifies the device to run the model on (cpu or cuda).

Advanced Usage

For advanced users, the detect_car.py script offers additional options to customize your detection pipeline:

• Save detection results to a specific directory.
• Toggle between different models or weight files.
• Modify input processing techniques or adjust the output format.

Documentation

For more detailed information, including how to extend the package or contribute to its development, please visit our official documentation.

Contributing

We welcome contributions to OurCustomPkg! If you have ideas for new features, enhancements, or bug fixes, please open an issue or submit a pull request on our GitHub repository.

License

This project is licensed under the MIT License. You can view the full license here.

Acknowledgments

• YOLOv7: Our package is built upon the innovative YOLOv7 model, which has significantly advanced the field of real-time object detection.
• MediaPipe: MediaPipe's hand tracking technology has enabled us to create a more interactive and user-friendly detection experience.

Contact

For any questions, feedback, or support, feel free to reach out to the authors:

• Brandon: [e-mail]
• Moshiur: [e-mail]

---

Thank you for using OurCustomPkg! We hope it serves your toy car detection needs and inspires further innovation in your projects.