VisioGrid

Table of Contents

• Project Introduction
• Main Features
• Tech Stack
• Installation and Running
• Usage

Project Introduction

VisioGrid is a heterogeneous distributed computing platform developed in Rust, focusing on image recognition. The project aims to establish an efficient distributed system for large-scale image recognition processing. By parallel processing tasks in a multi-agent environment, VisioGrid enhances the efficiency and speed of image recognition.

Main Features

• High Efficiency: Thanks to Tokio's asynchronous runtime and efficient I/O processing, VisioGrid can quickly handle large amounts of image data.
• Stability: Through real-time monitoring of agents and proactive error handling, the system runs stably, avoiding program crashes.
• Image Recognition: Supports multiple image recognition models, flexibly adapting to different application scenarios.

Tech Stack

• Rust: Fully utilizes Rust's high performance and safety features to ensure efficient system operation.
• Actix Web: Provides a user-friendly web management interface for easy system management and monitoring.
• Tokio: Uses Tokio's asynchronous runtime for performance optimization, improving I/O processing efficiency.
• GStreamer: Used for handling media streams and content, supporting efficient processing of multimedia data.
• Ultralytics: Integrates Ultralytics for image recognition, supporting multiple advanced image recognition models.

Installation and Running

System Requirements

• Nvidia GPU: This program can only run on machines equipped with Nvidia GPUs.

Compile from Source Code

Management Node

1. Clone the Repository

   git clone https://github.com/DaLaw2/VisioGrid
   cd VisioGrid

2. Install Required Packages

   bash build/ManagementDepend.sh

3. Compile the Project

   cargo build --release --package management

4. Edit Configuration File

   vim management.toml

5. Run the Node

   cargo run --package management --release

Agent Node

1. Clone the Repository

   git clone https://github.com/DaLaw2/VisioGrid
   cd VisioGrid

2. Install Required Packages

   bash build/AgentDepend.sh

3. Activate Virtual Environment and Install Dependencies

   python3 -m venv AgentVenv
   source AgentVenv/bin/activate
   pip3 install -r build/requirements.txt

4. Compile the Project

   cargo build --release --package management

5. Edit Configuration File

   vim agent.toml

6. Run the Node

   cargo run --package agent --release

Using Docker

VisioGrid provides Docker containers that include all necessary dependencies, eliminating the need for manual installation.

1. Create Docker Network

   docker network create VisioGrid

2. Build the Management Node Container

   cd build
   docker build -t management -f ManagementDockerfile

3. Run the Management Node Container

   docker run -it --rm --gpus all --network VisioGrid -p 8080:8080 management

4. Build the Agent Container

   cd build
   docker build -t agent -f AgentDockerfile

5. Run the Agent Container

   docker run -it --rm --gpus all --network VisioGrid agent

Usage

Access http://localhost:8080 via a web browser to view the management interface. The management interface provides system monitoring and management functions, allowing users to view system status, add agent nodes, check task progress, etc.

• The Home page provides an overview of the system, including system information, system load, and system logs.

  

• The Agents page displays all connected agent nodes. Clicking on an agent reveals more detailed information.

  

• The Inference page allows you to create a new inference task. After selecting a model and image, you can submit it.

  

• The Tasks page lists all tasks that are in progress, successful, or failed. You can view the execution status of each subtask.

  

• The Configuration page provides system configuration options, allowing users to modify system settings.