Agtech - Real-time Computer Vision and Object Detection

Overview

This repository contains Python scripts for various real-time computer vision tasks, inspired by the complex visual system of the mantis shrimp. The project aims to mimic the mantis shrimp's advanced vision capabilities, such as detecting subtle details in images that may be camouflaged or hidden.

The scripts in this repository include:

• Real-time camera input handling
• Interactive color picking for object detection
• Image processing techniques to detect and track objects
• Visualization and user interaction with detected objects

Table of Contents

• Installation
• Scripts Overview
• Usage
• Example Outputs
• License
• Contribution

---

Installation

Requirements:

Before running the scripts, ensure you have Python 3.x installed. You will also need the following libraries:

• OpenCV for video capture and image processing
• NumPy for numerical operations
• Matplotlib for visualizations
• Scikit-Image for image resizing
• Torch for tensor operations (for smoothing filters and processing)

Install the dependencies using pip:

pip install opencv-python numpy matplotlib scikit-image torch

Scripts Overview

mantisclickobject.py

Allows users to click on a live camera feed to pick colors and detect objects based on the selected color. It uses OpenCV for video capture and Matplotlib for visualizing the interaction.

autocamera.py

Automatically selects available camera indices and starts capturing video from the first available camera.

motion.py

Detects motion in a video feed, useful for object tracking or surveillance applications.

mantisbutton.py

Adds functionality for button interactions in the GUI (Graphical User Interface) to control processes during live video analysis.

mantiscolorpick.py

Provides an interactive GUI to pick specific colors in a camera feed for use in detecting similar colors in the image.

mantisshrimplive.py

A live feed script that simulates the vision abilities of the mantis shrimp, applying real-time filters to enhance colors or detect specific features in the video stream.

---

Usage

Running the Scripts

mantisclickobject.py

Run the script to open the camera feed and pick colors for object detection:

python mantisclickobject.py

Color Selection

Click on any part of the live video feed to pick a color.
The algorithm highlights pixels matching the selected color and suppresses others.

License

This repository is licensed under a proprietary license. Unauthorized copying, distribution, modification, or public display of this software is strictly prohibited without prior written permission from the owner. For more information, please refer to the LICENSE file or contact the repository owner.